Materials and methods 
2.1 Ethical approval
This double-blinded, balanced, randomised, crossover control trial was granted a favourable ethics opinion by the NHS, Berkshire B, Research Ethics Committee (21/SC/0378) and the Health Research Authority. This study conforms to the standards of the Declaration of Helsinki and the study protocol was preregistered on the ClinicalTrials.gov website (ID no. NCT05342012).
2.2 Participants
We recruited people with a clinical diagnosis of T2DM, as defined by the World Health Organisation.  Participants were ≥ 18 years of age and able to provide fully informed written consent. Participants were excluded if they had a body mass index (BMI) > 35, severe claustrophobia, currently smoked (or have stopped within 3 months), used proton pump inhibitors or phosphodiesterase inhibitors (or did not wish to stop using them for the study), or had any other serious medical condition or device that may interfere with data quality and/or participant safety during MRI (e.g. aneurysm clip, cardiac pacemaker/defibrillator, artificial implants, infusion pumps, metallic foreign bodies, surgical screws or similar, permanent piercings). In total, 15 participants (3 Female, 12 Male) were recruited from primary care (i.e. GP practices), local database of volunteers and via media releases (Table 1.) and their flow through the trial is presented in Figure 1. 
Participants were randomised (random.org) with concealed allocation into one of two experimental conditions to begin in either a control arm of nitrate depleted beetroot juice (placebo) or a nitrate rich beetroot juice (BJ). Participants and outcome assessors were blinded to treatment groups throughout. The placebo appears and tastes identical to BJ and, as previously reported, both have similar antioxidants and polyphenol content [38]. Final unblinding of all participants was performed after the creation of a locked analysis data sheet. Data were collected for BAT fat fraction, activation, and plasma [nitrate] and [nitrite]. The primary outcome was the change in BAT fat fraction. An a priori sample size calculation was performed to estimate the required n. For 80% power we required 13 people in order to detect a 1 standard deviation change in BAT fat fraction [39]. We aimed to recruit 15 people in order to account for 15% dropout rate.     

2.3 Pre-experimental tests 
Participants were provided with a participant information sheet ≥ 48 h prior to providing written informed consent. Participants were fully briefed about the study and provided with the opportunity to ask any questions prior to enrolment. A standard medical history and clinical exam were performed including; electrocardiogram (ECG), height, body mass, capillary blood samples from the fingertip and blood pressure. Blood was analysed in house for glycated haemoglobin (HbA1c) and lipid profiles (AfinionTM 2 Analyzer, Abbott, Abbott Park, Illinois, United States). Results from the 12-lead ECG were examined by a clinician alongside anything in the medical history that demonstrated questionable responses and/or contraindications for MRI prior to scanning. Participants were provided with the opportunity to be familiarised with the MRI scanning procedure to assist in screening for claustrophobia prior to commencing the study. 

At the end of visit 1 (consent and screening), the correct number of placebo or BJ bottles were placed in opaque bags for participants to take home. Each participant’s study number was used to randomly allocate treatment order. Participants were instructed when to take the juice and reminded that mouthwash was prohibited for the duration of the study, as it disrupts the oral microbiome responsible for nitrite conversion [6]. 

2.4 Protocol and Outcome measures 
Participants were instructed to ingest 140 mL·day-1 of BJ (delivers ~ 12.4 mmol·L1 of inorganic nitrate Beet It, James Whites Drinks Ltd.) or 140 mL·day-1 of placebo juice (0.1 mmol·L1) in a randomised order for 14-days prior to visit 2 and 3 and take the last 140 mL dose 1.5 hours before arriving at the laboratory. A minimum 7-day washout period was implemented between visits 2 and 3 in line with our other nitrate trials which have 7 days to be sufficient [10, 40-43]. 

The primary outcomes were assessed in the Bournemouth Gateway Building, Bournemouth, UK in July 2022 to January 2023. While the study period was six months, participants completed visits one and two within four weeks, with the minimum 7-day washout. Participants rested in ambient conditions for 10 minutes, following which BP was measured and, subsequently, a ~ 12 mL venous blood sample drawn from the antecubital fossa into a EDTA K2 (BD) and lithium heparin vacutainers (BD), for determination of plasma [nitrate] and [nitrite]. Participants were familiarised with the environment with 30-60 mins in the stable laboratory (22.8 ± 0.8°C, 51.4 ± 9.7% RH) before commencing infrared thermal imaging (described below). Following this, participants wore a cooling jacket for 1 hour. A water-bath (Optima TX150, Grant Instruments Cambridge Ltd, Royston, UK) with cooling coil (C1GR, Grant Instruments Cambridge Ltd, Royston, UK) was pre-filled and temperature maintained at 8.1 ± 1.2ºC and was used with a water perfused (~15 m of tubing) jacket (Cool Flow® Adjustable Cooling Vest System, Polar Products, Ohio, USA). Water temperature was monitored-using a thermistor inserted into the bath, and-ambient-temperature measured using a wet-bulb-globe thermometer (WBGT; HT200, Extech Instruments, Knoxville, Tennessee). Skin-thermistors (UU-VS, Grant Instruments Cambridge Ltd, Royston, UK), connected to a data logger (SQ2040 Squirrel Data Logger, Grant Instruments Ltd, Royston, UK) were attached to the chest (under cooling jacket and vest but not in contact with tubing), arm, thigh, and calf to calculate mean skin temperature (Mean Tsk) [44]. [34]. The thermal image was repeated pre- and post-cooling on both visits (see details below) and in addition a post-cooling MRI scan (3T MRI, Magnetom Lumina, Siemens Heathcare Ltd, Camberley, UK) was commenced within 2-3 mins of completion of the cooling protocol.  

MRI Imaging of Supraclavicular BAT
MRI of the supraclavicular fat depot was performed to quantify BAT activity. Image slices with 2 mm slice thickness and in-plane resolution of 1 x 1 mm were acquired using multi-point Dixon sequence (TR=15 ms, 10 echoes, TE1=1.23 ms, delta TE=1.23 ms) and body matrix coil after anatomical localization. The images had anatomical coverage of the neck, supraclavicular region, and the apices of the lung.
 

IRT Imaging of Supraclavicular BAT
For acquisition of thermal images taken from dry skin in the supraclavicular and sternal region perpendicular to the camera at 1 m distance (emissivity 0.95), each participant was seated in an upright position (wearing shorts and cotton singlet covering chest skin temperature site but not obstructing IRT measurement) on a chair, with their head in a neutral position looking straight ahead in a room where temperature was kept constant (temperature: 22.8 ± 0.8°C, relative humidity: 51.4 ± 9.8%). Respiratory-measures of pulmonary oxygen uptake (V ̇O2), carbon dioxide output (V ̇CO2) and minute ventilation (V ̇E) were recorded continuously (Cortex Metalyser 3B-R2, Cortex Medical, Seattle, Washington, USA, calibrated with a two-point calibration [ambient air; calibration gas, 5% CO2, 15% O2; with single use turbine flow checked volume measure]) to ensure that any temperature increases were a result of non-shivering thermogenesis; as significant changes in these variables is indicative of shivering thermogenesis [45].

Biochemical analysis 
Venous blood samples were placed in a centrifuge and spun at 4,000 g, 4°C for 10 min immediately following collection. Once spun, the plasma was pipetted into aliquots with a link anonymised code. The samples were then placed in a -80°C freezer until subsequent analysis. Samples were analysed for [nitrate] and [nitrite] using a Sievers nitric oxide analyser (Sievers NOA 280i, Analytix Ltd, Durham, UK) with our established technique [10, 38, 41, 42].

2.5 Data analysis 
MRI analysis 
The multi-echo complex data obtained using multi-point Dixon sequence were reconstructed using a multi-scale graph-cut algorithm [46]. Regions of interest (ROIs) were drawn within the left supraclavicular fat depots on the fat fraction image. A lower threshold of 40% and upper threshold of 85% fat fraction were used to exclude the muscle, bone marrow, and white adipose tissue that could have been inadvertently included within the ROI. The proton density fat fraction (PDFF) of BAT was quantified as the mean fat fraction within the ROIs. Recent developments in the non-invasive measurement of BAT using MRI [34, 39] conclude that fat fractions from the BAT regions can be reliably measured using MRI in comparison to PET related measures. 
IRT analysis 
Following ~30 minutes adjustment to the indoor ambient conditions (22.8±0.8°C, 51.4±9.7% RH) (outdoor conditions July-January 0.4-25.5°C) BAT activation was determined by the difference in skin temperature between the SCV fosse region and the sternal control region [37, 47, 48]. Temperature immediately post-cooling was assessed for the following variables: a) absolute SCV temperatures, and b) SCV – control values. 
Statistical analysis 

The distribution of data was assessed using descriptive methods (skewness, outliers, and distribution plots) and inferential statistics (Shapiro-Wilk test). Where normal distribution was violated, non-parametric analyses were performed. Paired t-tests were completed to determine differences in dependent variables between Beetroot Juice and Placebo interventions. Two-way analysis of variance (ANOVA) was completed to determine differences from baseline to end of cooling between interventions. Data are presented as mean (SD) unless otherwise stated. Statistical analysis was performed on GraphPad Prism (version 8) (GraphPad Software, Boston, MA) and statistical difference was accepted as two-tailed P < 0.05.