Rapid Characterization of Microalgae and Microalgae Mixtures Using Matrix-Assisted Laser Desorption Ionization Time-Of-Flight Mass Spectrometry (MALDI-TOF MS)

Current molecular methods to characterize microalgae are time-intensive and expensive. Matrix Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) may represent a rapid and economical alternative approach. The objectives of this study were to determine whether MALDI-TOF MS can be used to: 1) differentiate microalgae at the species and strain levels and 2) characterize simple microalgal mixtures. A common protein extraction sample preparation method was used to facilitate rapid mass spectrometry-based analysis of 31 microalgae. Each yielded spectra containing between 6 and 56 peaks in the m/z 2,000 to 20,000 range. The taxonomic resolution of this approach appeared higher than that of 18S rDNA sequence analysis. For example, two strains of Scenedesmus acutus differed only by two 18S rDNA nucleotides, but yielded distinct MALDI-TOF mass spectra. Mixtures of two and three microalgae yielded relatively complex spectra that contained peaks associated with members of each mixture. Interestingly, though, mixture-specific peaks were observed at m/z 11,048 and 11,230. Our results suggest that MALDI-TOF MS affords rapid characterization of individual microalgae and simple microalgal mixtures.


Introduction
Microalgae have received considerable attention in science and industry as they can be cultivated and harvested for many products and co-products including biofuels and nutraceuticals [1]. Microalgae have different growth rates which are affected by a range of environmental factors such as nutrient availability and temperature. Those environmental factors need to be controlled in order to generate product, especially in large-scale biomass production [2]; however, the environmentally-exposed open pond system model leaves microalgae cultures susceptible to contamination by undesired microalgae that can out-compete the original microalga for resources, which can negatively affect production [2] [3]. This shift in microalgae species can go unnoticed if the species are phenotypically similar. As a result, microalgae in mass-production systems need to be monitored regularly for contamination to avoid a decrease in productivity and catastrophic culture crashes.
Conventional techniques for microalgae identification include morphological analysis using bright field light microscopy and electron microscopy [4]. Complementary molecular techniques include multilocus sequence typing (MLST) [5], repetitive sequence-based polymerase chain reaction (rep-PCR) [6], 18S rDNA analysis [7], and pulsed-field gel electrophoresis (PFGE) [8]. In many instances, the use of these techniques requires amounts of time, labor, and resources that are impractical [9] for monitoring the health of microalgae ponds in near real-time.
Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is a technique that has been shown capable of rapidly and reliably characterizing bacteria at the genus, species, and in some cases, strain levels [10] and is becoming more routine in use [11]. Most often, this is achieved by comparing mass spectra (i.e., fingerprints) acquired from crude protein extracts of unknown microorganisms to reference spectra in databases [12]. Furthermore, studies have shown that MALDI-TOF MS-based fingerprint methods may afford greater taxonomic resolution than traditional molecular techniques [9][10] [13]. In addition to bacteria, MALDI-TOF MS has also been used to characterize fungi [14][15][16][17], viruses [18], and more recently to a considerably lesser extent, microalgae [4][5] [19] [20]. Nicolau and colleagues [4] obtained spectra of diatoms using MALDI-TOF MS and observed that culture age affected mass spectra. Von Bergen et al. [21] used MALDI-TOF MS to characterize five pathogenic species of Prototheca, and Wirth et al. [22] [23] showed that optimization of downstream analyses such as self-organizing mapping (SOM portrait analysis) of spectra allowed MALDI-TOF MS to discriminate between harmless and pathogenic Prototheca species. Most recently, Emami et al. [20] obtained greater taxonomic resolution during characterization of 31 strains of Dunaliella sp. with MALDI TOF MS than with internal transcribed spacer (ITS) sequence analysis. Each of these studies suggests that MALDI-TOF MS has promise as a tool for the rapid characterization of diverse, economically-relevant microalgae [24] [25].
To further explore the ability of MALDI-TOF MS to characterize microalgae, we focused on 31 algae representing 12 species. The specific objectives of this study were to determine whether MALDI-TOF MS can be used: 1) for species-level differentiation of economically-relevant algae; 2) for strain-level characterization; and 3) to characterize simple mixtures of microalgae. A common protein extraction sample preparation method was used. Sequence (18S rDNA) analysis was performed on all microalgae to confirm their identity and to compare the taxonomic resolution afforded by this traditional approach to a MALDI-based approach. Finally, two model mixture systems containing two and three microalgae were examined. Our results suggest that MALDI-TOF MS affords rapid: 1) characterization of a diverse collection of microalgae, 2) discrimination between multiple strains within a single species, and 3) characterization of simple mixtures.

Microalgae Cultivation
Thirty-one microalgae representing 10 genera and 12 species were provided by the Arizona Center for Algae Technology and Innovation (AzCATI; http://www.AzCATI.com) ( Table 1). Specifically, six of the genera were freshwater species (Chlamydomonas, Chlorella, Parachlorella, Chromochloris, Desmodesmus and Scenedesmus), and five were marine species (Dunaliella, Chlorella, Tetraselmis, Nannochloropsis, and Porphyridium). Five mL of BG-11 [26] for freshwater strains or F/2 [27] for marine strains were inoculated with a single colony of microalgae growing axenically on petri plates aseptically in a laminar flow hood. Microalgae samples were grown in 15 mL screwcap tubes with 75 μmol/m 2 /s of cool white fluorescent lighting at 20°C for 3 weeks prior to analysis with MALDI-TOF MS.

Sample Preparation for MALDI-TOF MS
A common, previously described protein extraction procedure was used as the basis for the sample preparation method used here [28]. One mL of cells at an optical density of 750 nm (OD 750 ) between 0.15 and 0.3 were washed with sterile milliQ-H 2 O (mQ-H 2 O) and then inactivated for 1 hour in 300 μL mQ-H 2 O and 900 μL absolute ethanol. Samples were then centrifuged at 10,000 x g for two minutes at room temperature. The supernatant was decanted, and the cells were resuspended in 1 ml mQ-H 2 O, centrifuged at 10,000 x g for two minutes once more, and the supernatant was again decanted. FA and ACN were added to the resulting pellet. Equal volumes of FA and ACN were added in volumes necessary to normalize to an initial culture OD 750 = 0.8. Pellets were vortexed vigorously. The samples were then centrifuged at 17,000 xg for five minutes at room temperature, and the supernatant was collected and used immediately for MALDI analysis.

Mass Spectra Acquisition
A Bruker Microflex LRF MALDI-TOF MS (Bruker Daltonics) was used to acquire mass spectra. The spectrometer was equipped with a 337 nm nitrogen laser and controlled using Flex-Control software (version 3.0; Bruker Daltonics). Mass spectra in the m/z 2,000 to 20,000 range were collected automatically in the positive linear mode. Ion source 1 was set to 20 kV, and ion source 2 was set to 18.15 kV with the lens set to 9.05 kV. Spectra for each sample were generated from 500 laser shots acquired in five 100 shot bursts. The laser frequency was set to 10 Hz. Spectra from each of the 100 shot bursts were included only if the following parameters were met: a base peak (i.e., the peak with the greatest intensity) signal-to-noise ratio (S:N) of 2 or greater, a peak width of 10 m/z, a minimum intensity threshold of 100, and a maximal number of peaks of 500. Three replicate MALDI mass spectra were obtained per algal strain. Peak smoothing was performed using the Savitzky-Golay algorithm. Baseline subtraction was performed using the TopHat algorithm. Calibration of the mass spectrometer was performed using a protein calibrant mixture containing the proteins listed above. Peaks were identified using FlexAnalysis 3.0 software (Bruker Daltonics) and then transferred to a Microsoft Excel spreadsheet in which the average mass range and base peak signal-to-noise ratios were calculated to assess spectrum quality. Peaks were considered different if they varied by more than +/-2 m/z.

MALDI-TOF MS Data Analysis
Additional analysis of spectra was performed using BioNumerics (v. 7; Applied Maths, Austin, TX, USA). Composite spectra (i.e., summary spectra) were created using data from all replicates to represent each of the 31 microalgae. A similarity threshold of 65% was used to ensure representation of each replicate spectrum in the summary spectra. Pseudo-gels were constructed to visualize the MALDI profiles for each microalga. Similarity was quantified using the Pearson correlation coefficient, and a dendrogram was generated by using the UPGMA method.

Characterization of mixtures of microalgae using MALDI-TOF MS
Two model mixture systems were constructed using samples composed of two or three microalgae cultures. The OD 750 for each microalga was adjusted to 0.3 before mixing so that the microalgae would be represented equally in the mixture. The first mixed culture contained two microalgae, Chlorella vulgaris UTEX 395 and Scenedesmus acutus LRB-AP 401. To construct the mixture, 500 μL of each culture were added to a microcentrifuge tube to yield a 1 mL solution. The second mixed culture contained three microalgae: C. vulgaris UTEX 395, S. acutus LRB-AP 401, and Chlorella sorokiniana UTEX 1230. The 1 mL mixture was constructed using 333 μL of each microalga. Samples were prepared for MALDI analysis, and mass spectra were acquired using the procedures described above. Peak lists of the three individual samples and the two mixtures were compared. Each peak was included in the peak list only if it was present in all replicates. Peak matching was performed to identify prominent peaks attributed to individual isolates and peaks that were mixture-specific.
Mass ranges and peak numbers varied among the spectra of microalgae examined here. Masses in the spectrum of C. vulgaris UTEX 395 ranged from m/z 2,044 to 10,685. Comparatively, spectra of P. purpureum LRB-OH 6101, Tetraselmis sp. CBS 15-2610, and C. reinhardtii CC 849 exhibited narrower mass ranges (as low as m/z 2,005 and up to m/z 8,701 with Tetraselmis sp.). Spectra of N. oceanica IMET-1 had the broadest mass range (m/z 2,087 to 13,265). Numbers of peaks for the samples described here ranged from 6 for Chromochloris zofingeinsis UTEX 32 up to 56 peaks for Nannochloropsis granulata CCMP 525. These results are comparable to the work of Lee et al. [7], who used MALDI-TOF MS to characterize Nannochloropsis granulata, Chlorella sp., and Dunaliella sp. While spectra shown here are not identical to those described previously (e.g., Lee et al. [7] reported a broader mass range with Chlorella as well as a prominent peak near m/z 8,700 not observed in our spectra), both our work and that of Lee et al. [7] suggests that MALDI affords rapid and clear differentiation of diverse microalgae.
Moving beyond species-level characterization, we examined the capability of MALDI to characterize microalgae at the strain-level. The works of Murugaiyan [23], von Bergen [21], and Wirth [22] using members of the genus Prototheca support the ability of MALDI to distinguish between strains of microalgae within the same species. Our data suggest that strain-level differentiation of members of the genus Chlorella is feasible. Mass spectra of three Chlorella strains (C. vulgaris UTEX 395, C. vulgaris UTEX 259, and C. vulgaris LRB-AZ 1201) are clearly distinct (Fig 2A-2C). Spectra of all three microalgae exhibited similar mass ranges of m/z 2,182-10,685; 2,167-10,660; and 2,182-9,283, respectively; however, the spectra contained different base peaks at m/z 6,422; 2,637; and 2,517, respectively. These distinct spectra may be explained, in part, by the work of Gerken et al., who demonstrated previously that while the 18S rDNA sequences between 11 C. vulgaris strains were over 99% similar, the sensitivity of each strain to specific enzymes was remarkably distinct, indicating a highly variable cell wall composition among the various strains [29].
Similar to our results with C. vulgaris, spectra of three Nannochloropsis strains (N. oceanica CCMP 531, N. oceanica IMET-1, and N. oceanica CCAP 849/10) appeared to afford strainlevel differentiation (Fig 2D-2F). Spectra of all three of these strains contained a characteristic peak near m/z 8,378. The spectrum of strain CCAP 849/10 contained a different base peak (m/z 4,240) than the other two N. oceanica strains. Additional differences in peaks among spectra of these strains were observed (Fig 2D-2F).
Most recently, Emami et al. [20] have reported results similar to ours in which MALDI-TOF MS appeared to afford greater taxonomic resolution in microalgae than gene sequence-based methods. In particular, they were able to differentiate strains of Dunaliella. Similar to our work, they also used a mass range of m/z 2,000 to 20,000. Our work is comparable in size (i.e., number of isolates analyzed), but broader (i.e., focus beyond a single genus) in taxonomy compared to the work of Emami et al. Interestingly, Emami et al. reported that whole cell-based sample preparation was necessary to yield useful spectra [20]. Our results, however, suggest that a relatively common, protein extraction-based approach to sample preparation is sufficient to produce MALDI spectra of microalgae that yield species-and strain-level characterization. While spectra we report here and those reported by Emami [20] for C. vulgaris are not identical, prominent peaks below m/z 3,000 are observed in spectra produced by both groups. Differences between spectra are likely related to different sample preparations, different strains, and differences in life stages used in each study.

Comparison of 18S rDNA Sequence and MALDI-TOF Data
Differences observed in the spectra summarized above were reflected in the cluster analysis of spectra of all 31 microalgae examined here (Fig 3A). Spectra of microalgae clearly separated at the species level. Separation of microalgae at the strain-level was also observed (Fig 3A).
We also performed 18S rDNA sequence analysis and compared it to the MALDI-TOF MS data. The 18S rDNA-based dendrogram (Fig 3B) included five clades corresponding to five classes of microalgae represented in our collection. As expected, members of the same genus and species clustered together; however, at the strain level, the 18S rDNA sequence data did not afford clear separation of Nannochloropsis salina strains and Scenedesmus acutus strains. N. salina CCMP1776 and CCMP537 sequences had no differences in 18S nucleotide sequences; S. acutus LRB-AP 401 and LRP-AZ 414 differed by only 2 nucleotides. Additional DNA sequencing data of regions such as ITS-2 would be required to clearly differentiate these strains. In contrast, the MS-based dendrogram (Fig 3A) clearly separated nearly all strains examined including strains of N. salina, S. acutus, C. vulgaris, C. reinhardtii, N. oceanica, N. gaditana, and P. purpureum. The dendrogram based on the 18S rDNA data demonstrated a much higher degree of taxonomic organization (i.e., members of the same class clustered together) compared to the MS-based dendrogram as has been reported previously [7]. Similar to our results, Lee et al. [7] reported intermixing of taxonomically similar microalgae at the class-level in an MS-based dendrogram. Differences between MALDI-and 18S rDNA sequence-based dendrograms are reflective of the facts that: 1) 18S rDNA dendrograms are based only on gene sequence data, while MALDI dendrograms and spectra contain proteome-level, gene expression-based data and 2) different clustering algorithms are routinely employed with each type of data (i.e., gene sequence data are typically clustered using the neighbor-joining algorithm [5], while MALDI spectra are often clustered using the Unweighted Pair Group Method with Arithmetic Average (UPGMA) algorithm [9]).

Mixture Analysis
Rapid detection of contaminating microalgae and deleterious community shifts are important during outdoor pond cultivation of microalgae biomass. For this reason, we attempted to use MALDI to characterize simple mixtures of microalgae. As has been reported frequently with MALDI analysis of bacterial mixtures, spectra of mixtures of microalgae contained many peaks originating from the individual microalgae composing the mixture [31][32][33][34][35][36]. The first mixture contained C. vulgaris UTEX 395 and S. acutus LRB-AP 401. Six prominent peaks from these two individual microalgae were observed in the spectrum of this mixture (Table 2; Fig 4). The second mixture contained C. vulgaris UTEX 395, S. acutus LRB-AP 401, and C. sorokiniana UTEX 1230. Spectra from this mixture contained eight prominent peaks found in the spectra of the constituent three individual microalgae.
Not all peaks observed in spectra of individual microalgae were observed in the mixture spectra. Peaks present in the spectrum of C. vulgaris UTEX 395 dominated both mixture spectra ( Table 2; Fig 4). As previously postulated [31], ion suppression may account for underrepresentation of individual microalgae in the spectra of mixtures. Ion suppression results when one analyte suppresses appearance of another in a mass spectrum due to: 1) the suppressing ion being present at a higher concentration than the suppressed ion and/or 2) the suppressed ion does not ionize as efficiently as the visible ion. We adjusted the OD 750 of each microalgae sample to 0.3 before constructing the mixtures, but the C. vulgaris peaks remained the most prominent in the mixture spectra. It is possible that the C. vulgaris yielded more readily ionized proteins compared to S. acutus, but further work is warranted to further clarify mechanisms of peak suppression in microalgal mixtures.
Interestingly, spectra of both mixtures exhibited unique peaks at m/z 6,481; 11,048; and 11230. These three peaks appear to be mixture-specific as they do not appear in the spectra of the individual microalgae constituents. Similar results have been reported previously with bacterial mixtures [31], in which two mixture-specific peaks were observed in a mixed culture of E. coli and S. Typhimurium. The origin of these peaks and the mechanism of their formation is not clear, but may result from interactions between proteins (e.g., enzymes) associated with the individual cultures. Alternatively, interspecies interactions between the algae may have induced expression of proteins represented by these peaks. In either case, these mixture-specific peaks may provide information that is useful in the rapid characterization of algal mixtures and/or identification of contamination of algal cultures.

Conclusions
Our results suggest that MALDI-TOF MS represents a rapid and effective alternative to conventional methods of characterizing microalgae. To our knowledge, this is the first report of the use of MALDI to characterize mixtures of microalgae (polycultures), which are gaining popularity within the microalgae production industry. The taxonomic resolution of this rapid approach appears superior to conventional gene-sequencing based methods, as has been reported recently with Dunaliella [20]. Mixture-specific peaks were observed and may serve as biomarkers of contamination that allow producers to rapidly detect contamination events. Accordingly, MALDI-TOF MS has potential as a more rapid and economical means of monitoring the health and productivity of microalgae culture systems. For this reason, our current efforts include development of sample preparation and data analysis workflows that facilitate rapid analysis of more complex microalgal mixtures, including those that result from contamination events and predator introduction.