Participant recruitment and data collection were carried out in the Nutritional Physiology Research Unit at the University of Exeter between August 2021 and May 2022. This study was approved by the Sport and Health Sciences Ethics Committee of the University of Exeter (21–07-14-A-01) in accordance with the Declaration of Helsinki and is registered at ClinicalTrials.Gov (NCT05401591). Experimental procedures, potential risks, and the purpose of the study were explained to the participants prior to obtaining informed written consent. Before their inclusion in the study, participants attended a screening session which consisted of assessments of body mass, height, blood pressure, body composition (BodPod, Life Measurement, Inc.), and the completion of a routine medical screening questionnaire. Participants were enrolled in the study after being deemed healthy based on these results, with the exclusion criteria of a BMI below 18.5 or above 30 kg·m⁻²; high blood pressure (>140/90 mmHg); underlying health conditions which might affect metabolic function; known allergies or intolerance to drink ingredients; or regular smokers.

Fifteen young, healthy participants (age, 21 ± 1 years; BMI, 24 ± 1 kg·m²; male:female, 8:7) were initially recruited and consented to take part in the present study. Three participants withdrew from the study due to adverse effects (vomiting) following the ingestion of a microalgae drink, one participant became uncontactable after their first visit and therefore did not complete all conditions, and one further participant completed the study but due to an incomplete data set were not used for analysis. Therefore, excluding the above mentioned, 10 young, healthy participants (age, 21 ± 1 years; BMI, 25 ± 1 kg·m²; male:female, 7:3) completed this study and comprise the dataset. Participants’ characteristics (n = 10) are presented in Table 1.

Participants were monitored by a researcher throughout each of their study test days for signs of adverse events. If a participant had an adverse event, the test day was immediately ended to manage participant comfort and avoid the adverse event influencing data collected during the postprandial period. Where an adverse event occurred and the participant had not completed all other test conditions, the entire data set was excluded (i.e., the three participants who withdrew). Adverse events (vomiting) occurred in an additional four participants who were included in the analysis. The onset of these adverse events occurred after >210 min of the test day had been completed, missing data analysis was applied for the remaining data points, accounting for <4% of data sets. For three of these participants the adverse event was during their final visit, and for one participant it occurred on their third visit, however, they were able to return for the final condition. A flowchart of the described timings of adverse events for all participants (including those withdrawn) are detailed in Supplementary material 1. Overall, two adverse events occurred following SPR ingestion, five following WCC, and one following SCC.

In a randomised, double-blind, crossover and counterbalanced (for order) design, participants completed four experimental test days. Randomisation was performed by an independent researcher using a computerised randomiser. During each visit, participants ingested a beverage containing 20 g protein derived from milk (MLK), spirulina (SPR), or chlorella in a whole cell (WCC) or split cell (SCC) form. Arterialised-venous blood samples were collected in the postabsorptive state and at regular time intervals over a 5 h postprandial period to assess circulating amino acid, glucose, insulin, and uric acid concentrations. Indirect calorimetry and visual analogue scales (VAS) were used at regular intervals to determine energy expenditure, and subjective appetite and palatability scores, respectively. Test days for a given participant were separated by at least 7 days to allow for complete digestion, absorption, and metabolism of the test proteins, and return to habitual dietary habits.

Prior to taking part in the trial, participants completed a 3-day food record to assess habitual dietary intake on two weekdays and one weekend day. Records were analysed for energy and macronutrient intake using dedicated software (Nutritics Ltd.). Data are presented in Table 1. Participants were instructed to refrain from vigorous physical activity and alcohol consumption for 24 h before each test day. The evening before each test day, participants were provided with a standardised meal. Participants were allowed to choose between 2 standardised meals (energy, kJ/ carbohydrate, g/ fat, g/ protein, g: Meal 1: 1916/49/20/19, Meal 2: 2235/64/17/27) and had to adhere to the same choice prior to each experimental test day.

An overview of the experimental protocol is shown in Figure 1. On all test days, participants reported to the laboratory at 08:00 h after a > 12 h overnight fast and were asked to rest semi-supine on a hospital bed. A Teflon cannula was inserted retrograde into a dorsal hand vein, attached to a 0.9% saline infusion for patency (infusion rate 20 mL·h⁻¹), and placed in a heated hand unit (55°C) for subsequent arterialised-venous blood sampling (47). Participants completed a visual analogue scale (VAS) to assess subjective feelings of appetite. These 100 mm paper-based scales included questions regarding hunger, satisfaction, fullness and prospective food consumption (48). Afterwards, indirect calorimetry (Cortex Metalyzer 2R gas analyser; Cortex) measurements were collected for 20 min to record resting metabolic rate. A postabsorptive blood sample was then collected (t = 0). Participants consumed one of the four protein beverages in a randomised and counterbalanced (for order) manner. They were instructed to consume the beverages within 5 min, and drink completion indicated the start of the postprandial period. Consumption of the beverage was followed by a 5 h postprandial period during which 11 further arterialised venous blood samples were collected at t = 15, 30, 45, 60, 90, 120, 150, 180, 210, 240 and 300 min, while subjects remained in a semi-supine position throughout. Further appetite VAS recordings were collected at t = 5, 60, 120, 180, 240 and 300 min, with additional beverage palatability questions assessing taste, aftertaste, texture and overall palatability at t = 5 min. The same researcher analysed VAS scales each time to minimise discrepancies, with collected data used to calculate individual and composite appetite and palatability scores as previously reported (25, 49) (see section “Calculations”). Resting metabolic rates were determined hourly at t = 45, 105, 165, 225, 285 for 15-min intervals. Analysis of the last 5 min of each period was used to obtain average V̇ O₂ and V̇ CO₂ values to determine energy expenditure following test drink ingestion using the modified Weir equation (50) (see section “Calculations”).

Spirulina; SPR (Bulk™, United Kingdom), whole cell chlorella; WCC (eChlorial®, France), and split cell chlorella; SCC (Golden Greens Organic Ltd., United Kingdom) were purchased in dried powder form from commercial suppliers. Specific growing conditions of the biomasses were not provided by the suppliers. Obtainable pertinent information was as follows: WCC was cultivated in glass tubes and the cell walls kept intact (43), whereas SCC was grown in pools and after drying their cell walls broken (44). The dried microalgae powders were used as received by commercial suppliers with no additional processing performed before beverage preparation (see below). Thorough validation of cell wall disruption was not possible; however, a qualitative optical microscopy approach was taken and provided below (see “Qualitative microscopy”). All protein sources were independently analysed (Premier Analytical Services, United Kingdom) for energy, macronutrient and amino acid compositions, with the details presented in Table 2. Protein content was calculated as nitrogen (N) × 6.25 (N determined via the Kjehdal method by Premier Analytical Services, United Kingdom). While the presence of non-protein nitrogen-containing factors potentially introduces a small amount of error, we consider this to be in line with what is typically accepted within the food industry (51) and similar studies within the literature (24, 25, 52). To ensure isonitrogenous and isovolumetric conditions could be achieved across beverages, the milk protein beverage consisted of commercially obtained instant full cream milk powder (Nestlé UK Ltd., United Kingdom) dissolved in skimmed milk (Tesco Stores Ltd., United Kingdom).

Protein beverages were prepared the evening before test days by adding the amount of powder required (43, 32, 36, and 34 g for MLK, SPR, WCC and SCC, respectively) to provide 20 g protein to 200 mL water, or 200 mL skimmed milk (MLK condition). Maltodextrin (Myprotein, United Kingdom) was added to make each drink up to a total of 75 g of carbohydrates. Fifteen mL of artificial energy-free vanilla flavouring (Jordan’s Skinny Mixes, UK) and 10 mL of green energy-free food colouring (Tesco Stores Ltd., United Kingdom) were added for blinding purposes (making all drinks a dark green colour) and mixed for approximately 2 min using a food blender. Following drink consumption by the participant, an additional 100 mL was added to ‘rinse’ the bottle and ensure that all protein had been consumed, making a total fluid volume of 400 mL consumed by participants on each occasion. Double-blinding of the drinks was achieved by having a different researcher to the one performing the experiment prepare the drinks in a metal, non-transparent bottle ready for consumption. Drinks were refrigerated at 4°C until use. Following ingestion of each drink, volunteers were asked to identify which condition they thought they had received which was noted down without feedback. The overall success rate for volunteers correctly identifying the condition was 60%. Individual successful identification rates were as follows: SPR, 40%; MLK, 100%; SCC, 40%; WCC, 60%, implying partially successful blinding, with the exception of MLK.

Eight mL of arterialised venous blood were collected into a syringe at each sampling point. For each blood sample a 20 μL plastic capillary was filled and immediately analysed for blood glucose concentrations (Biosen C-Line GP+, EKF diagnostics). The remaining whole blood was split equally into serum separator tubes (BD Vacutainer SST II tubes, BD Diagnostics), and lithium heparin-containing tubes (BD Vacutainer LH; BD Diagnostics). The blood tubes were handled according to the manufacturer’s instructions (serum separator tubes were left at room temperature for 30 min to clot prior to centrifugation, lithium-heparin tubes were centrifuged immediately). Following centrifugation (1,300 ×g at 4°C for 10 min) the resulting blood serum and blood plasma supernatants were aliquoted and stored at −80°C until subsequent analysis.

Serum uric acid concentrations were measured enzymatically via colorimetry on a Cobas 8,000 automated analyser (Roche Diagnostics) as described previously (37). Serum insulin concentrations were determined using a commercially available ELISA assay kit (DRG Insulin ELISA, EIA-2935, DRG International Inc., United States) following the manufacturer’s instructions.

Plasma concentrations of alanine, glutamic acid, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tyrosine, and valine were determined by GC–MS with electron impact ionisation (Agilent) as previously described (25, 52). Briefly, to prepare samples for GC–MS, 10 μL of 2 mM norleucine were added as an internal standard to 500 μL of plasma and deproteinised on ice with 500 μL of 15% ^w/_v 5-sulfosalcylic acid. Samples were then vortexed and centrifuged at 4000 ×g for 10 min at 4°C. The supernatant was then loaded onto cation-exchange columns. Columns were filled with ddH₂O, followed by 6 mL 0.5 M acetic acid and then washed 5 more times with ddH₂O. Amino acids were then eluted from the columns with 2 mL of 6 M ammonium hydroxide. The eluate was dried using a Speed-Vac at 60°C and then derivatised via the addition of 50 μL of N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide1% tertbutyl-dimethylchlorosilane and 50 μL of acetonitrile, followed by heating at 95°C for 45 min. Derivatized samples were subsequently analysed on a GC–MS as previously described (25, 52).

Optical brightfield microscopy was used to qualitatively observe any differences in the commercial suppliers described whole cell- and split cell-chlorella preparations. See Supplementary material 2 for details.

Modified Weir equation (50):

Appetite score (49):

Based on previous research using a similar randomised cross-over design (53, 54), sample size was calculated with differences in postprandial plasma essential amino acid (EAA) incremental area under the curve (iAUC) between protein sources as the primary outcome measure. A sample size of 12 participants was calculated with a power of 80% and a significance level of 0.05 to detect a relevant difference in EAA iAUC between protein sources.

For each parameter, all four conditions were compared within the same statistical test and analysed by a two-way ANOVA with repeated measures (with condition and time as factors). In the event of a significant main effect, Tukey’s multiple comparisons tests were applied to locate individual differences. Or, where the iAUC was calculated, a one-way ANOVA was performed to detect significant effects of condition. If a significant main effect was detected, Tukey’s multiple comparisons tests were again applied to locate individual differences. Statistical significance was set at p < 0.05. N = 10 unless stated otherwise, with missing data handled using expected-maximisation algorithm. All statistical analysis were performed by using GraphPad Prism version 10 (GraphPad Software). All data are expressed as means with their standard errors.

Postabsorptive (i.e., t = 0) and postprandial plasma total amino acid (TAA), essential amino acid (EAA), and non-essential amino acid (NEAA) concentrations following protein drink ingestion are shown in Figures 2A,C,E. From similar postabsorptive levels, plasma TAA, EAA and NEAA concentrations all increased following drink ingestion (time effects; all p < 0.001), although to different extents between groups (time x condition interactions; all p < 0.001).

Postabsorptive plasma TAA concentrations (Figure 2A) for MLK, SPR, WCC and SCC were 1,544 ± 105, 1,514 ± 127, 1,666 ± 144 and 1,507 ± 109 μmol·L⁻¹, respectively (p = 0.38). Post drink ingestion, plasma TAA concentrations reached peaks of 2,361 ± 121, 2,344 ± 88, 2002 ± 99 and 2093 ± 126 μmol·L⁻¹ after 72 ± 26, 78 ± 8, 57 ± 13 and 81 ± 7 min, in MLK, SPR, WCC and SCC, respectively. For SCC there was a ‘double peak’ around 45 min and again at 105 min. These data translated to different postprandial TAA availabilities between conditions, reflected by postprandial iAUC (above the baseline value) varying significantly (Figure 2B, p = 0.002) depending on protein source (MLK; 86.6 ± 17.8, SPR; 84.9 ± 12.5, WCC; −4.1 ± 21.7, SCC; 55.7 ± 11.3 mmol·L⁻¹). Specifically, TAA iAUC following SPR and MLK ingestion were similar (p = 0.99), but both were higher than WCC (p < 0.05) but not SCC (p > 0.05). SCC showed a trend for higher TAA iAUC compared with WCC (p = 0.09).

A similar pattern was seen with plasma EAA (Figure 2C), where from similar postabsorptive plasma concentrations (MLK 698 ± 41; SPR 681 ± 48; WCC 744 ± 54; SCC 665 ± 42 μmol·L⁻¹, p = 0.39), peak EAA concentrations differed (p < 0.05) between MLK and SPR (1,171 ± 64 and 1,139 ± 52 μmol·L⁻¹, respectively) vs. WCC and SCC (881 ± 48 and 894 ± 48 μmol·L⁻¹, respectively). This translated to different postprandial EAA availabilities, reflected by iAUC varying significantly (Figure 2D, p = 0.0002) depending on protein source (MLK; 45.0 ± 10.0, SPR; 42.7 ± 8.6, WCC; −11.8 ± 9.9, SCC; 14.3 ± 4.9 mmol·L⁻¹). Specifically, EAA iAUC following SPR and MLK ingestion were similar (p = 0.99), both were higher than WCC (p < 0.001), but not SCC (p > 0.05). No differences were detected between chlorella conditions (p = 0.19).

For the NEAA (Figure 2E), postabsorptive plasma concentrations were similar across conditions (MLK 846 ± 63; SPR 832 ± 79; WCC 922 ± 90; SCC 842 ± 66 μmol·L⁻¹, p = 0.64) and the peak NEAA concentrations did not differ (1,221 ± 73, 1,250 ± 59, 1,156 ± 57 and 1,224 ± 87 μmol·L⁻¹ for MLK, SPR, WCC and SCC, respectively, p > 0.05). The NEAA postprandial iAUC (Figure 2F) varied depending on protein source (MLK; 41, 0.7 ± 8.6, SPR; 42.1 ± 8.9, WCC; 7.6 ± 12.5, SCC; 41.3 ± 7.1 mmol·L⁻¹, p = 0.046); however, multiple comparisons did not detect individual differences between drinks.

The iAUC for WCC was negative when calculated for the entire 5 h postprandial period for TAA and EAA (not NEAA). However, when the postprandial period was split into 0–2.5 h and 2.5–5 h periods the WCC iAUC was positive for the first half of the postprandial period. For 0–2.5 h significant differences were detected (p > 0.05) for TAA between WCC compared with all other conditions (MLK, SPR and SCC). This is illustrated in Supplementary material 3.

Data showing the postprandial plasma responses of each measured individual amino acid are displayed in Supplementary material 4. All individual amino acid plasma concentrations reported a time effect (p < 0.05), except glutamic acid (p = 0.19). Differences between conditions (p < 0.05) were only detected for isoleucine, leucine, methionine and tyrosine. Time × condition interactions were reported for all the measured individual amino acid plasma concentrations, except histidine (p = 0.25) and lysine (p = 0.13).

From similar postabsorptive values, blood glucose concentrations (Figure 3A) increased after drink ingestion (time effect; p < 0.001) but with no differences between conditions (condition and time × condition interaction; p = 0.20 and 0.99, respectively). Postprandial blood glucose iAUC did not differ between conditions (Figure 3B, p = 0.91).

Postabsorptive and postprandial serum insulin concentrations are presented in Figure 3C. From similar postabsorptive concentrations (MLK, 11 ± 2; SPR, 14 ± 2; WCC, 14 ± 3; SCC, 15 ± 2 mU·L⁻¹; p = 0.74), serum insulin concentrations increased following drink ingestion (time effect; p < 0.001), but with no differences between conditions (condition and time × condition interaction; p = 0.89 and 0.40, respectively). Postprandial insulin iAUC also did not differ between conditions (Figure 3D, p = 0.89).

Resting whole-body energy expenditure in the postabsorptive state and at regular intervals during the 5 h postprandial period is displayed in Figure 4. Postabsorptive resting energy expenditure was equivalent between conditions (MLK, 61 ± 5 [1,475 ± 113]; SPR 66 ± 3 [1,585 ± 74]; WCC, 67 ± 3 [1,609 ± 77]; SCC, 66 ± 5 [1,588 ± 110] kcal· h⁻¹ [kcal·24 h⁻¹]; p = 0.78). Drink ingestion resulted in a significant increase in energy expenditure from postabsorptive values (time effect; p < 0.001), and for the entire duration of the postprandial period for MLK (p < 0.05), and for 2 h following the ingestion of WCC and SCC (p < 0.05). However, no differences between conditions were detected (condition and time × condition interaction; p = 0.94 and 0.10, respectively).

Participants’ subjective composite appetite scores during the experimental trials are presented in Figure 5A. Individual ratings of hunger, fullness, prospective food intake, and desire to eat are presented in Supplementary material 5. Following drink ingestion, main effects of time for all appetite variables were observed (all p < 0.001), but no condition effect (p = 0.83) or time x condition effects (p = 0.90).

A subjective ‘overall palatability’ score for each condition is presented in Figure 5B, with values recorded of MLK 72 ± 6 mm, SPR 30 ± 7 mm, WCC 26 ± 7 mm and SCC 36 ± 5 mm. Palatability scores for taste, smell, aftertaste, and texture are presented in Supplementary material 6. Scores of each palatability variable, including overall palatability, differed between protein sources (condition effect; all p < 0.05). Across all variables, except texture, MLK scored higher than all microalgae drinks (p < 0.02), and there were no differences between the microalgae drinks (p > 0.05). For texture, the only difference detected was MLK scoring higher than WCC (p < 0.03).

Serum uric acid concentrations in the postabsorptive state and over the 5 h postprandial period are shown in Figure 6. Postabsorptive serum uric acid concentrations were similar in all conditions (MLK, 314 ± 15; SPR, 313 ± 22; WCC, 316 ± 22; SCC, 286 ± 17; p = 0.68). There was a significant time (p < 0.0001) and time × condition interaction (p < 0.0001), but no significant effect of the condition alone (p = 0.21). In the MLK condition, serum uric acid concentrations decreased steadily across the postprandial period, whereas for the microalgae conditions (SPR, WCC and SCC) serum uric acid concentrations increased by 1 h and remained elevated for the duration of the postprandial period, such that at 300 min post drink ingestion, differences were detected between MLK and all microalgae conditions (p < 0.05; with no individual differences postprandially between microalgae conditions).

See Supplementary material 2 for optical brightfield microscopy images of whole cell- and split cell-chlorella. Observationally, fewer cell clusters were observed in the split cell compared with the whole cell-chlorella preparation.