Natural killer cell subset count and antigen‐stimulated activation in response to exhaustive running following adaptation to a ketogenic diet

Abstract We investigated the effect of a 31‐day ketogenic diet (KD) compared with a habitual, carbohydrate (CHO)‐based diet on total circulating natural killer (NK) CD3−CD56+, dim and bright subset count, and antigen‐stimulated CD3−CD56+ cell activation (CD69+) in response to exhaustive running. In a randomised, repeated‐measures, cross‐over study, eight trained, male endurance athletes ingested a 31‐day low‐CHO KD or their habitual diet (HD). On day 31, participants ran to exhaustion at 70% V˙O2max (∼3.5–4 h, ∼45–50 km). A low‐CHO (<10 g) meal was ingested prior to the KD trial, with fat ingested during exercise. A high‐CHO (2 g kg−1) meal was ingested prior to the HD trial, with CHO (∼55 g h−1) ingested during exercise. Venous blood samples were collected at pre‐exercise, post‐exercise and 1 h post‐exercise. The KD amplified the classical exercise‐induced biphasic CD3−CD56+ cell response by increasing the post‐exercise counts (P = 0.0004), which appeared to be underpinned by the cytotoxic CD3−CD56dim subset (main effect of time point, P < 0.0001). The KD had no effect on NK cells’ expression of CD69 or their geometric mean fluorescence intensity of CD69 expression, either for unstimulated or for antigen‐stimulated NK cells (all P > 0.05). In conclusion, adaptation to a KD may alter the number of circulating NK cells but not their ability to activate to an antigenic challenge.

immune responses do not reflect immune cells of the innate immune system, which act as our first line of defence against pathogens.
Natural killer (NK) cells of the innate immune system can respond differently to T-cells following identical dietary exercise interventions (Fletcher & Bishop, 2012) and are sensitive to changes in CHO-availability (Henson et al., 1999). NK cells are large granular lymphocytes that play a critical role in the first line of defence against pathogens and neoplastic cells (Caligiuri, 2008). They are identified by the cell surface marker CD3 − CD56 + and can be divided into subsets (Caligiuri, 2008). Generally, CD56 bright NK cells, which comprise ∼10% of total circulating NK cells, appear to be important for coordinating 'cross-talk' between the acquired and innate arms of the immune system due to their capacity to produce cytokines, whereas CD56 dim NK cells are more cytotoxic and can lyse cells without antibody recognition (Caligiuri, 2008). CD56 dim NK cell mobilisation also appears sensitive to adrenergic stimulation (Dimitrov et al., 2010), suggesting the greater adrenergic response from a KD could alter exercise-induced NK cell trafficking patterns from secondary lymphoid to peripheral tissues via the blood and therefore amplifying the biphasic NK response.
Further, upon activation, NK cells express CD69 (i.e., early activation marker) (Borrego et al., 1999;Werfel et al., 1997). CD69 expression is important for NK cell function, including roles in cell proliferation, cytolytic activity and cytokine secretion (Borrego et al., 1999;Werfel et al., 1997). Only ∼10% of unstimulated peripheral blood NK cells express CD69, which increases several-fold upon antigen stimulation (Werfel et al., 1997). However, the relationship between NK cell CD69 expression, exercise and CHO-availability is not well-understood. Despite NK cell function typically being impaired by increased cortisol concentrations (Muscari et al., 2022), the relationship between corticosteroids and NK cell CD69 expression is less clear, with little work currently conducted in this area.
Further investigation into the effect of a KD on immune function is warranted, particularly in populations engaging in exhaustive exercise lasting several hours, who may be at an increased risk of illness. Therefore, we continued our investigation into the effects of a KD on immunity through examining circulating NK cells in response to exhaustive running in trained male endurance athletes. We hypothesised that a KD would increase total NK cells and the CD56 dim NK cell subset following exercise but suppress CD3 − CD56 + NK cell activation (expression of CD69) following in vitro multi-antigen stimulation during recovery. This study is part of a series assessing the effects of a KD on performance and health (Maunder et al., 2021;Shaw et al., 2019Shaw et al., , 2021. • What is the main finding and its importance?

Ethics approval
The KD amplified the biphasic exercise-induced NK cell response due to a greater mobilisation of the cytotoxic CD56 dim subset but did not alter NK cell CD69 expression. The KD appears to modulate exercise-induced circulating NK cell mobilisation and egress, but not antigen-stimulated circulating NK cell activation.
was not registered in a database. All participants signed an informed consent form prior to participation.

Participants
Participants were required to have been: (1) male; (2) habitually consuming a mixed diet for >12 months; (3) weight stable for >1 month; (4) running >50 km per week; and (5) completed a marathon in <3.5 h within the previous 6 months. Only males were included to avoid the potential effect of the menstrual cycle and various types of contraception on immune and physiological indices.
Participants were excluded if they: (1) reported a history of fat-or keto-adaptation (i.e., intentionally reduced dietary CHO intake <150 g per day for several weeks in order to amplify exercising fat oxidation); (2) previously ingested exogenous ketone supplements; (3) were currently or recently injured; (4) experienced moderate-to-severe gastrointestinal symptoms or illness within the previous 4 weeks; (5) had a history of irritable bowel syndrome; (6) habitually smoked; or (7) had been ingesting dietary supplements or medications known to affect immune function within the previous 2 weeks, with the exception of caffeine, protein and CHO supplements. All characteristics were acquired by self-report. Eight eligible healthy, trained, male endurance athletes (two marathoners, four ultra-marathoners and two longdistance triathletes) participated in the study. Participants were fully informed of the study's rationale and possible risks of the experimental procedures before providing their written consent.

Experimental overview
A detailed overview of this study's design, experimental trials, dietary intervention and monitoring, and training has been published previously (Shaw et al., 2019

Blood sampling and analysis
Venous blood samples were collected at pre-exercise, post-exercise and 1h post-exercise into an a 6 ml sodium heparin and a 6 ml K 2 EDTA Vacutainer (BD, Franklin Lakes, NJ, USA) with the participants seated in an upright position. Due to the rapid exercise-induced lymphocyte kinetics (Rooney et al., 2018), all blood samples were collected within 1 min of the specified time points. The cannula was flushed with 3-4 ml of saline every 30 min to maintain patency. Each serum vacutainer was left to clot for 30 min prior to centrifugation at 1500 g for 10 min at 4 • C.
After this, samples were separated into two 1.5 ml aliquots to be stored at -80

Haematological analysis
K 2 EDTA treated whole-blood was used to determine red blood cell, haematocrit, haemoglobin and total lymphocyte concentration haemoglobin and Hct is haematocrit (Dill & Costill, 1974).

Multi-antigen-stimulated whole-blood cultures
Blood samples from the sodium heparin vacutainer were used to set up both unstimulated and stimulated whole-blood cultures in Falcon 12 × 75 mm polystyrene tubes with caps (BD Biosciences, Auckland, New Zealand). Zero or 20 μl of the multi-antigen working vaccine (1:100) was added to 200 μl of heparinised whole blood, giving a final stimulant concentration of 1:1000, before being incubated for 24 h at 37 • C and 5% CO 2 (Midi 40 CO 2 incubator, Thermo Fisher Scientific, Waltham, MA, USA). The stimulant was a commercially available multi-antigen vaccine containing diphtheria toxoid, tetanus toxoid, Bordetella pertussis antigens, hepatitis B surface antigen, poliovirus and Haemophilus influenzae, and was used to elicit a recall immune response. The 1:1000 stimulant dilution was chosen based on previous titration work/pilot work performed in our lab to illicit optimum CD69 expression on NK cells.

Natural Killer cell count and CD69 expression
After 24

Data analysis
One

Circulating CD3 − CD56 + natural killer cell count
There was a diet × time point interaction for circulating CD3 − CD56 + NK cell count (P = 0.004; Figure 2). Cell counts were similar between dietary conditions at pre-exercise (P = 0.661). In the KD, cell counts increased from pre-to post-exercise (P < 0.0001) and were 46% higher compared with the HD at post-exercise (P = 0.0004). Cell counts declined by ∼60% in the KD from post-exercise to 1-h post-exercise (P < 0.0001) to below pre-exercise counts (P = 0.040). Cell counts also declined by ∼30% from post-exercise to 1 h post-exercise in the HD (P = 0.040) and remained similar to the KD at 1-h post-exercise (P = 0.350).

F I G U R E 2
Circulating CD3 − CD56 + cell counts in the habitual and ketogenic diet conditions. Values are presented as raw mean and individual responses for n = 7 participants. Diet × time point interaction: higher in the ketogenic diet (KD) compared with the habitual diet (HD) condition for time point (*P < 0.05); higher at post-exercise compared with pre-exercise in the KD ( a P < 0.05); lower at 1 h post-compared with post-and pre-exercise in the KD ( b P < 0.05); and lower at 1 h post-exercise compared with post-exercise in the HD ( c P < 0.05).

DISCUSSION
We examined the effect of a 31-day KD compared with a habitual CHO-based diet on circulating CD3 − CD56 + NK cells, the CD56 bright and CD56 dim subset counts, and in vitro unstimulated and antigenstimulated CD3 − CD56 + NK cell activation in response to exhaustive moderate-intensity running (∼3.5-4 h, ∼45-50 km). Our primary observations were partly in support of our hypotheses and include: (1) the KD augmented the transient increase of circulating CD3 − CD56 + cell counts immediately post-exercise, which declined to below pre-exercise counts at 1-h post-exercise; (2) the CD56 dim subset appeared the predominant subset shifting circulating NK cell counts; (3) circulating unstimulated CD3 − CD56 + cells expressing CD69 were lower 1-h post-exercise compared with post-exercise, but not when antigen-stimulated; and (4) there was no change in the antigen-stimulated CD69 expression (per cell expression of activation) in response to exercise or diet. These findings suggest that the KD amplifies the biphasic NK cell responses to prolonged exhaustive exercise, which appeared to be underpinned by the CD56 dim subset, but does not influence the ability of NK cells to activate to an antigenic challenge during recovery.
Adaptation to a KD shifts substrate utilisation toward fat to reduce the contribution of the body's finite stores of CHO to energy production during endurance exercise (Shaw et al., 2020). This is a practice that has grown in popularity over recent years within the athletic population, with varied performance outcomes being demonstrated. We further polarised substrate availability and metabolism in our study by implementing acute pre-exercise and during-exercise fuelling techniques. As reported in our previous study, blood ketone body and cortisol concentration increased in the KD condition, while blood glucose concentration remained similar between KD and HD conditions , suggesting blood glucose concentration was unlikely to be a cause of immune differences between dietary conditions. Despite not measuring the catecholamine response, we have previously demonstrated that adaptation to a KD increased autonomic stress (suggestive of an increased catecholamine response) indirectly via reductions in resting heart rate variability (Maunder et al., 2021), which is consistent with other studies investigating KDs (Langfort et al., 1996(Langfort et al., , 1997. Considering these marked physiological differences between conditions, alterations in circulating NK cell number and function in response to exercise were expected following adaptation to the KD. Exercise elicited a biphasic circulating NK cell count response, with an increase from pre-to post-exercise, followed by a decline during the 1-h recovery period. The trafficking of NK cells is a primary contributor to the classical exercise-induced lymphocytosis and lymphopenia, which we observed in our previous study . This was likely due to lymphocyte mobilisation from secondary lymphoid tissue, followed by redistribution to peripheral tissues (Gleeson & Bishop, 2005). A similar pattern occurred for the CD56 dim subset, whereas the CD56 bright subset was unaltered. This preferential mobilisation of the more cytotoxic CD56 dim subset is similar to other studies (Campbell et al., 2009;Timmons et al., 2006), whereas the CD56 bright subset, which have a predominant regulatory role, are relatively unaffected by exercise and typically recirculate between the blood and secondary lymphoid tissues (Cooper et al., 2001). Prolonged moderate-intensity exercise, compared with short high-intensity exercise, may also have less effect on the CD56 bright subset (Gannon et al., 1997); therefore, it is unsurprising that the circulating CD56 bright subset was unaltered in the present study. These findings highlight the need to report both circulating CD56 bright and CD56 dim subsets to better understand circulating NK cell function and mobilisation with exercise.
The KD amplified the NK cell count response to exercise. Despite the KD having no clear influence on the exercise-induced changes of the CD56 dim subset (P = 0.054), this cytotoxic cell subset seemed likely to underpin this effect, particularly due to having a larger response to exercise. Similarly, short-term low-CHO diets (Mitchell et al., 1998) and exercise without CHO supplementation (Henson et al., 1999) also appear to exacerbate the biphasic NK cell response to exercise. In our study, the increase in CD56 dim cells also appears higher when exercising without CHO supplementation (Timmons et al., 2006), which is likely due to a greater catecholamine response preferentially mobilising and redistributing cytotoxic lymphocytes via stimulation of β 2 -adrenergic receptors that are highly expressed on these cells (Dimitrov et al., 2010;Maisel et al., 1990). In contrast, post-exercise reductions in the CD56 dim subset have been down to be exacerbated with CHO supplementation (Timmons et al., 2006). This was potentially due to selective margination of the CD56 dim subset, which could be due to increased leukocyte adhesion to endothelium through upregulation of adhesion proteins (Morigi et al., 1998), with CD56 dim cells more rapidly adherent than CD56 bright cells (Vujanovic et al., 1993). In our F I G U R E 3 (a) CD56 bright subset percentage of total lymphocytes, (b) CD56 bright subset cell count, (c) CD56 dim subset percentage of total lymphocytes, and (d) CD56 dim subset cell count in the habitual and ketogenic diet conditions. Values are presented as raw mean and individual responses for n = 7 participants. Effect of time point: higher at post-compared with pre-exercise ( a P < 0.05); and lower at 1-h postcompared with post-exercise ( b P < 0.05).
study, the lack of differences in blood glucose concentration between conditions may explain why we did not observe this.
Only exercise influenced the number of unstimulated circulating NK cells expressing CD69, which declined during 1 h of recovery. However, following antigen stimulation, this effect was abrogated, suggesting the NK cells retained their ability to activate. The greater autonomic stress response in the KD condition (Langfort et al., 1996) was expected to increase the mobilisation of NK cells primed for effector functions (i.e., CD69 + ) into circulation following exercise; however, there were no effects of diet on the number of NK cells expressing CD69 or the intensity of CD69 expression. It is possible that despite increased circulating cytotoxic cells, alternative mechanisms masked potential effects of diet and exercise. For example, in our previous study , cortisol and antigen-stimulated whole blood IL-10 production were higher in the KD condition, which may have led to an immunosuppressive environment post-exercise. Nevertheless, higher cortisol concentrations (following a high dose of caffeine) have been reported to not alter the number of antigen-stimulated NK cells expressing CD69 in response to 90 min of strenuous exercise (Fletcher & Bishop, 2011).
The applicability and implications of our findings may be limited due to the characteristics of the sample population and methods employed. Endurance trained athletes that can maintain moderateintensity exercise for several hours are not representative of the general population, but were recruited for two primary reasons: (1) the nature of their training was more likely to supress immune function (Walsh et al., 2011); and (2) they are theoretically more metabolically endowed to gain an endurance benefit from adapting to a KD than their less trained counterparts. As such, it is difficult to ascertain if these same patterns of response would be seen amongst the general population undertaking less exhaustive exercise.
Females were not included in the current study, and as such we are unable to translate these findings; however, previous research suggests exercise-induced changes in circulating NK cell number and function can be consistent across sexes (Moyna, 1996 (Campbell et al., 2009), which can confer host protection. However, exercise-induced trafficking patterns of (activated) NK cells remain unknown.
In conclusion, we demonstrated that adaptation to a 31-day KD augments the biphasic circulating NK cell count response to exercise, which appeared to be due to a greater mobilisation and redistribution of the cytotoxic CD56 dim subset. However, the KD had no effect on circulating NK cells' ability to activate (CD69 + ) in response to in vitro antigen stimulation. The exercise protocol also lowered the number of unstimulated NK cells expressing CD69 1-h post-exercise, although this effect was abolished following in vitro antigen stimulation. Overall, the KD appears to uniquely modulate circulating counts of NK cells following prolonged exhaustive exercise but not their ability to activate to antigen stimulation. This suggests the KD has little effect on the functional capacity of this aspect of the innate arm of the F I G U R E 4 (a) Number of unstimulated CD3 − CD56 + cells expressing CD69, (b) unstimulated CD3 − CD56 + cells expressing CD69 as a percentage of total CD3 − CD56 + cells, (c) number of antigen-stimulated CD3 − CD56 + cells expressing CD69, and (d) antigen-stimulated CD3 − CD56 + cells expressing CD69 as a percentage of total CD3 − CD56 + cells in the habitual and ketogenic diet trials. Values are presented as raw mean and individual responses for n = 7 participants. Main effect of time point: lower at 1-h post-compared with post-exercise ( a P < 0.05).