Supplementary Materials for

Vitamin D regulates microbiome-dependent cancer immunity

Evangelos Giampazolias et al.

Corresponding authors: Evangelos Giampazolias, [email protected]; Caetano Reis e Sousa,

[email protected]

Science 384, 428 (2024)

DOI: 10.1126/science.adh7954

The PDF file includes:

Materials and Methods

Figs. S1 to S12
References

Other Supplementary Material for this manuscript includes the following:

Tables S1 to S3
MDAR Reproducibility Checklist
Supplementary Materials
Materials and Methods
Mice
Mice lacking Gc-globulin (Gc-/-) carrying the Gctm1.1(KOMP)Vlcg allele on a C57BL/6 background were
purchased from KOMP repository. Mice with inserted loxP site upstream of the exon 3 of Vdr
allele (Vdrtm2Ska) on a C57BL/6 background were purchased from The Australian Phenome Bank
at the Australian National University. VillinCre mice were crossed to mice carrying the Vdr-floxed
allele at the Francis Crick Institute to generate VdrDIEC strain that lacks VDR expression in gut
epithelium. Cgas-/- mice carrying the Cgastm1a(EUCOMM)Hmgu/IcsOrl allele on a C57BL/6 background
were imported from the MRC Human Immunology Unit, Weatherall Institute of Molecular
Medicine, University of Oxford, John Radcliffe Hospital as CRUK import. The above mice, as
well as Rag1-/-, Batf3-/-, Tap1-/-, Ifngr-/-, Ifnar-/-, Myd88-/-, Trif-/- and wildtype (WT) C57BL/6J were
bred and maintained in specific-pathogen free (SPF) conditions in the Biological Research
Facility (BRF) at The Francis Crick Institute. WT C57BL/6J (The Francis Crick Institute), C57BL/6J
(Charles River) and C57BL/JolaHsd (Envigo) are maintained at the CRUK Manchester Institute.
C57BL/6 germ free (GF) mice were bred and maintained at the at the Frederick National
Laboratory Gnotobiotics Facility for the National Cancer Institute (NCI), NIH. Mouse genotypes
were determined using quantitative PCR with specific probes designed for each gene
(Transnetyx, Cordova, TN).

Mice were used at 6 - 21 weeks of age for experiments. For tumor challenge, males and females
were used as we did not observe sexual dimorphism (not shown). However, in any one
experiment, mice were sex-matched and randomly assigned to treatment or control groups.
Mice of different genotypes were littermates and/or separately housed or co-housed in groups
for a minimum of 3 weeks before experiments as stated in the figure legends. Animal experiments
that were conducted in the UK were performed in accordance with national and institutional
guidelines for animal care and were approved by the Francis Crick Institute and CRUK
Manchester Institute Biological Resources Facility Strategic Oversight Committee (incorporating
the Animal Welfare and Ethical Review Body) and by the Home Office, UK. Animal experiments
that were conducted in US were approved by the Institutional Animal Care and Use Committee
of National Cancer Institute and were conducted in accordance with the IACUC guidelines and
the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Giampazolias et al 2

Diets
Mice at the Francis Crick Institute were fed with the commonly used standard chow diet (Teklad
Global 18% protein rodent diet, 2018S, Envigo) that contains standard VitD concentration (2
IU/g). Mice at the CRUK Manchester Institute were fed with the commonly used standard chow
diet (Teklad Global 18% protein rodent diet, 2018, Inotiv) that contains standard VitD
concentration (1.5 IU/g). Mice were fed for 3.5 weeks with modified diet to restrict or enhance
VitD levels (66). For this, VitDDeficient (VitD 0 IU/g, TD.89123, Teklad custom diet, Envigo), VitDHigh
(VitD 10 IU/g, TD.98234, Teklad custom diet, Envigo) or control VitDStandard (VitD 2 IU/g,
TD.110798, Teklad custom diet, Envigo; similar to Teklad Global diet at the Francis Crick
Institute) were used. VDRDIEC and littermate control mice were maintained on a VitDStandard+ diet
supplemented with 2% calcium, 1.25% phosphorus and 20% lactose (VitD 2 IU/g, TD.210810,
Teklad custom diet, Envigo) to mitigate the osteomalacia-like effects of abrogating VitD
responsiveness in gut epithelium (35). For experiments, some VDRDIEC and littermate control mice
groups were switched to VitDHigh+ diet (VitD 10 IU/g, TD.210811, Teklad custom diet, Envigo).

Administration of antibiotics
For depletion of microbiota, 0.5 g/L vancomycin (Fisher Scientific, AC296990050), 1 g/L
metronidazole (Sigma, M3761) or 1 g/L neomycin (Sigma, N6386) was provided in autoclaved
drinking water (changed twice per week) in the presence of sweeteners (Sweet'N Low, 93200)
starting 2 weeks before tumor implantation and continuing throughout the study.

Fecal matter transplantations

Freshly or snap frozen (stored at -80oC) stools of similar size were collected from donor mouse
strains. The stools were resuspended in PBS (1 fecal pellet/1mL PBS) and 200μl of the dissolved
fraction was administered via oral gavage on day -14 and day -12 to mice prior to tumor
inoculation on day 0. The recipient mice were deprived of food for 2 hours before and 30 minutes
after each fecal transplantation.

Microbial culture and administration to mice

Bacteroides fragilis (NCTC 9343; ATCC 25285) and Prevotella brevis GA 33 (ATCC 19188) were
procured from the American Type Culture Collection and cultured at 37°C under anaerobic
Giampazolias et al 3

conditions in a vinyl anaerobic chamber (COY) inflated with a gas mix of approximately 12%
carbon dioxide, 86% nitrogen and 2% hydrogen. Prior to the experiment, bacteria were pre-
cultured twice, once in modified Gifu anaerobic medium broth (mGAM, HyServe) and then in
brain heart infusion broth (BHI, Sigma-Aldrich) supplemented with 5µg/mL hemin and 0.5µg/mL
Vitamin K (BBL™ Vitamin K1-Hemin Solution, BD Biosciences). For generating aliquots to be
administered to mice, microbes were inoculated into a larger culture volume (500mL and 1 litre
supplemented BHI) and grown overnight to an optical density of approximately 2. Bacterial
cultures were harvested under anaerobic conditions by centrifuging at 4000 rpm for 20 min and
the pellet was washed twice with PBS. Microbial pellets were then adjusted to a cell density of
109 bacteria/200μL in PBS containing 25% glycerol, aliquoted in 200 µl aliquots and stored at -
80˚C. Viability and purity was tested by plating on mGAM agar plates and observing under a
microscope.
Microbial species or vehicle (25% glycerol in PBS) were administered via oral gavage on
day -14, day -12 and day -10 to mice under the hood, two weeks prior tumor inoculation (day 0).
The recipient mice were deprived of food for 2 hours before and 30 minutes after each gavage.

Quantification of bacteria by qPCR

Mouse fecal samples were collected 2 weeks after colonisation with B. fragilis or vehicle, snap
frozen in liquid nitrogen and stored at -80oC. Genomic DNA was isolated from fecal samples
using the QIAamp DNA Stool Mini Kit (Qiagen) following the manufacturer’s instructions.
Targeted qPCR was performed using the SYBR Green PCR Master Mix (ThermoFisher) on a
QuantStudio 7 Flex Real-Time PCR System (ThermoFisher). The following primers (oligo
sequence 5’-3’) and probes were used. B.fragilis 16rRNA gene forward primer:
TGATTCCGCATGGTTTCATT and reverse primer: CGACCCATAGAGCCTTCATC (71). Universal
16S rRNA forward primer: ACTCCTACGGGAGGCAGCAGT and reverse primer:
ATTACCGCGGCTGCTGGC (72). B. fragilis abundance was calculated relative to universal 16S
using 2-ΔΔCt method.

Tumor cell injections

Tumor cell lines (5555 BrafV600E (Caetano Reis e Sousa, The Francis Crick Institute (73)), MCA-
205 (George Kassiotis, The Francis Crick Institute) and MC-38 (ENH204-FP, The Francis Crick
Institute)) were grown in RPMI 1640 (Thermo Scientific-gibco, 31870025) containing 10% FCS
(Sigma, F0926-500ML), 2 mM glutamine / 100 units/mL penicillin / 100mg/mL streptomycin
Giampazolias et al 4

(Thermo Scientific-gibco, 10378016) 50mM 2-mercaptoethanol (Thermo Scientific-gibco,

31350010). All media and media supplements were from Life Technologies except for FCS
(Source Bioscience). Tumor cells were dissociated with 0.25% trypsin (Thermo Scientific-gibco,
15090046) and washed three times in PBS. The final cell pellet was resuspended and diluted in
endotoxin-free PBS (between 0.2 x 106 to 0.5 x 106 cells per 100μl) and injected s.c. in the shaved
right flank of each recipient mouse. Tumor growth was monitored every 2 to 3 days, and the
longest tumor diameter (l) and perpendicular width (w) were measured using digital Vernier
calipers; tumor volume was calculated using the formula (74): length x width2 / 2 and is expressed
as mm3. Mice that were poorly injected and did not subsequently develop tumors through the
course of the study were excluded from the analysis.

In vivo administration of immune-checkpoint blockade therapy

For immune-checkpoint therapy in vivo, anti-PD1 monoclonal antibody (clone RMP1-14,
BioXCell, BE0146) or rat IgG2a isotype control (clone 2A3, BioXCell, BE0089) was administered
i.p. at 200μg / 200mL PBS per mouse from day 3 post-tumor cell transplantation, every 3 days
up to a maximum of six doses. Alternatively, mice received anti-CTLA-4 monoclonal antibody
(clone 9D9, BioXCell, BP0164) or rat IgG2b isotype control (clone MPC-11, BioXCellBE0086)
50μg / 200mL i.p. on days 6, 9 and 12.

In vivo CD8 T cell depletion

For CD8+ T cell depletion, mice received 300μg / 200mL of anti-CD8 (clone 2.43, BioXCell,
BE0061) or rat IgG2b isotype control (clone LTF-2, BioXCell, BE0090) i.p. from 3 days prior to
inoculation of tumor cells and followed twice per week until the end of the experiment (days: 1,
4, 7, 10, 13, 16, 19, 22).

Analysis of immune populations in the tumor tissue

Tumors were excised at the indicated days after cell transplantation. Tumor weight of individual
tumors was determined using a microscale. For subsequent analysis by flow cytometry, tumors
were cut into pieces and digested with 200 U/mL collagenase IV (Worthington, LS004188) and
100mg/mL DNase I (Roche, 11284932001) for 60 min at 370C. Tissue was passed through a 70
μm cell strainer (Falcon), washed with FACS buffer (PBS with1% FCS and 2mM EDTA) and cells
were incubated with Fc block (CD16/32, clone 2.4G2, BD Biosciences, 553141) for 10 min in 4oC
before proceeding with antibody mediated staining.
Giampazolias et al 5

For the ex vivo analysis of T cells, a total of 5x106 isolated cells were treated with 50
ng/mL PMA (Sigma, 16561-29-8), 1 μg/mL ionomycin (Santa Cruz, 56092-82-1) and 1x brefeldin
A (eBioscience, 00-4506-51) in complete RPMI medium and incubated for 4 h in Corning 96-well
plates. Cell suspensions were stained with LIVE/DEAD Fixable Blue Dead Cell dye (Life
Technologies, L34962) according to manufacturer’s protocol and subsequently stained with the
following antibodies: BV785-CD45.2 (104, Biolegend, 109839, AB_2562604, 1:200 dilution),
APC/Cy7 TCRβ (H57-597, Biolegend, 109220, AB_893624, 1:100 dilution), BV605-CD8a (53-6.7,
Biolegend, 100744, AB_2562609, 1:200 dilution), PerCP/Cy5.5-CD4 (RM4-5, BD Biosciences,
553052, AB_394587, 1:200 dilution), APC-CD44 (IM7, BD Bioscience, 559250, AB_398661,
1:200 dilution). Cells were fixed and permeabilized using the Fixation/Permeabilisation buffer-
Foxp3 Kit (E-Biosciences, 00-5523-00) according to the manufacturer’s protocol. For
intracellular staining, samples were stained in permeabilization buffer using the following
antibodies. FITC-IFNγ (XMG1.2, BD Pharmingen, 554411, AB_395375, 1:100 dilution), Pacific
Blue-GZMB (GB11, Biolegend, 515408, AB_2562196, 1:200 dilution), PE-FOXP3 (FJK-16s, E-
Bioscience, 12-5773-82, AB_465936, 1:200 dilution). Quantification of total cell numbers by flow
cytometry was done using beads (Beckman Coulter). Samples were acquired on a Fortessa X20
B (BD Biosciences). Data were analysed using FlowJo software. After gating on single, live
CD45.2+ cells, immune cell populations were defined the following: CD4+ and CD8+ T cells (CD8a-
CD4+ and CD4-CD8a+ as a proportion of gated live CD45.2+TCRβ+), CD44+, IFNγ+ and Foxp3+
CD4+ T cells (CD44+, IFNγ+ or Foxp3+ as a proportion of gated live CD45.2+TCRβ+CD8a-CD4+),
CD44+, IFNγ+ CD8a+ T cells (CD44+ or IFNγ+ as a proportion of gated live CD45.2+TCRβ+CD4-
CD8a+).

Analysis of intestinal lamina propria lymphocytes and gut-associated lymphoid tissues

Mouse small intestine, cecum and colon were collected in IMDM (Thermo Scientific-gibco,
21980-082) containing 5% FCS. In the case of the small intestine, Peyer’s patches were
removed. Mouse intestines were cut longitudinally and carefully washed with PBS to remove
fecal particles. Mucus was carefully removed from the intestines by scraping with a scalpel
blade. Then, the intestines were vortexed (5 seconds) in ice-cold PBS and the supernatant was
discarded. Intestinal tissues were placed in Falcon tubes with pre-warmed IMDM containing 1%
FCS, 5 mM EDTA, 1X pen-strep (Thermo Scientific-gibco, 15140-122), 10 mM HEPES (Thermo
Scientific-gibco, 11360039), 1X glutamine (Thermo Scientific-gibco, 25030-081) and 2mM DTT
(Invitrogen, Y00147) and were incubated at 37°C for 20 min in shaking incubator to remove
Giampazolias et al 6

epithelial cells. Supernatant were removed and the tissues were washed with cold IMDM/PBS.
Small intestines were cut in pieces of 0.5cm length and were placed in Falcon tubes with pre-
warmed IMDM medium containing 1% FCS, 1X pen-strep, 10 mM HEPES, 1X glutamine, 1
mg/mL Collagenase VIII (Sigma, C2139-1G) and 33.3 μg/mL DNAse I (Roche, 11284932001) and
were incubated at 37°C for 10 min in a shaking incubator. Colon and cecum samples were cut
in pieces and were placed in Falcon tubes containing pre-warmed IMDM medium containing 1%
FCS, 1X pen-strep, 10 mM HEPES, 1X glutamine, 0.42 mg/mL Liberase (Roche, 05401020001)
and 33.3 μg/mL DNAse I (Roche, 11284932001) and were incubated at 37°C for 30 min in a
shaking incubator. FACS buffer was added to inhibit enzymatic activity and subsequently the
tissue suspensions were passed through 100μm cell strainer on top of a 50-mL tubes using a
syringe plunger to force any remaining tissue through the strainer. Cell suspensions were
centrifuged at 600g for 10 min, the supernatants were discarded and the cell pellets were
resuspended in 40% (v/v) Percoll diluted in IMDM (GE Healthcare,17-0891-01) and were
centrifuged at 450g for 8 min. Percoll was aspirated and the remaining cells were restimulated
with 50ng/mL PMA, 1 μg/mL ionomycin and 1x brefeldin A in complete RPMI medium and
incubated for 4 h in Corning 96-well plates.
For the ex vivo analysis of intestinal lamina propria lymphocytes, cell suspensions were
stained with LIVE/DEAD Fixable Blue Dead Cell dye (Life Technologies, L34962) according to
manufacturer’s protocol and subsequently stained with the following antibodies: BV785-CD45.2
(104, Biolegend, 109839, AB_2562604, 1:200 dilution), APC/Cy7 TCRβ (H57-597, Biolegend,
109220, AB_893624, 1:100 dilution), PE-Cy7-TCR-delta (GL3, Biolegend, 118124,
AB_11203530, 1:100 dilution), BV605-CD8a (53-6.7, Biolegend, 100744, AB_2562609, 1:200
dilution), CD4 BV785 (RM4-5, Biolegend, 100551, AB_11218992, 1:200 dilution). Cells were fixed
and permeabilized using the Fixation/Permeabilisation buffer-Foxp3 Kit (E-Biosciences, 00-
5523-00) according to the manufacturer’s protocol. For intracellular staining the samples were
stained in permeabilization buffer using the following antibodies: PE-FOXP3 (FJK-16s, E-
Bioscience, 12-5773-82, AB_465936, 1:200 dilution), Pacific blue-IL17A (TC11-18H10.1,
Biolegend, 506918, AB_893545, 1:100 dilution), APC-IL22 (IL22JOP, Invitrogen, 17-7222-82,
AB_10597583, 1:100 dilution), FITC-IFNγ (XMG1.2, BD Pharmingen, 554411, AB_395375, 1:100
dilution). Quantification of total cell numbers by flow cytometry was done using beads (Beckman
Coulter). Samples were acquired on a Fortessa X20 B (BD Biosciences). Data were analyzed
using FlowJo software. After gating on single, live CD45.2+ cells, immune cell populations were
defined the following: CD4+ and CD8+ T cells (CD4+ and CD8a+ as a proportion of gated live
Giampazolias et al 7

CD45.2+TCRdelta- TCRβ+), IFNγ+, IL22+ and IL17A+ CD4+ T cells (IFNγ+, IL22+ or IL17A+ as a
proportion of gated live CD45.2+ TCRdelta-TCRβ+CD4+), CD8+IFNγ+ T cells (IFNγ+ as a proportion
of gated live CD45.2+ TCRdelta-TCRβ+CD8a+), TCRγδ+ T cells (gated on live CD45.2+TCRβ-
TCRdelta+), IFNγ+, IL22+ and IL17A+ TCRγδ+cells (IFNγ+, IL22+ or IL17A+ as a proportion of gated
live CD45.2+TCRβ-TCRdelta+), ILCs (gated on live CD45.2+TCRβ-TCRdelta-), IFNγ+, IL22+ and
IL17A+ ILCs (IFNγ+, IL22+ or IL17A+ as a proportion of gated live CD45.2+ TCRβ-TCRdelta-).
Small intestine and colon samples were prepared using the “Swiss roll” technique, fixed
in 4% paraformaldehyde and stained with hematoxylin and eosin. Images were obtained with
10x magnification using Slide Scanner (Zeiss Axio Scan.Z1). Gut-associated lymphoid tissues
(GALT) were semi-quantitatively assessed blindly by two board-certified veterinaries using a gut-
associated lymphoid tissue scoring system, with a score from 0 – 2 (0 = absent to minimal; 1 =
moderate hyperplasia; 2 = marked hyperplasia). Morphologically, a marked GALT score was
given when there were abundant and/or single hyperplastic lymphoid follicles containing a wide
germinal center with apoptotic and mitotic figures.

Gut permeability assay

Mice were fasted for 4 hours prior to oral administration with 12mg of freshly prepared 4 kDa
fluorescein isothiocyanate (FITC)-dextran (Sigma-Aldrich) in PBS. As a positive control of
increased gut permeability, WT mice were treated with 2% DSS (MP biomedicals) in drinking
water for 5 days prior to FITC-dextran administration. Blood was collected 30min after oral
gavage via the tail vein. At 3 hours post injection, mice were culled, and blood was collected
from the heart. As a negative control, blood was sampled from a WT mouse that was not injected
with FITC-dextran. Blood was centrifuged at 1,000g for 10 min at 4oC, and serum was diluted
1:2 in PBS. 100μl of diluted serum and standards (0-200 mg/mL) were plated in duplicates, and
fluorescence was measured using a Tecan spectrophotometer (Tecan Group Ltd.) at an
excitation wavelength of 485nm and emission wavelength of 535nm.

Generation of bone marrow chimeras

CD45.1 or CD45.1.2 WT mice were exposed to two doses of 6.6Gy γ-irradiation with a 5-hour
interval. The following day the mice received i.v. 2x106 BM cells from CD45.2 WT or Myd88-/-
mouse donors in 200μL PBS. The mice were placed on acidified drinking water 1 week prior to
and 5 weeks following irradiation. Subsequently, reconstituted chimeric mice were returned to
non-acidified drinking water for 3 weeks before FT from WT or Gc-/- mice (week 8 post-BM
Giampazolias et al 8

transplantation). Chimeric mice were challenged s.c. with tumors at two weeks post-fecal
transplantation as described above.
Mice were bled once by venipuncture of a superficial blood vessel to assess bone marrow
chimerism in blood cells before they received the first fecal transplant as follows. Mouse blood
was mixed with EDTA (0.5M) and Fc block at 4oC for 10 min and subsequently stained with
LIVE/DEAD Fixable Blue Dead Cell dye (Life Technologies, L34962, 1:500 dilution) according to
manufacturer’s protocol and subsequently stained with the following antibodies: APC-CD45.1
(A20, Invitrogen, 17-0453-82, AB_469398, 1:400 dilution), PE Cy7-CD45.2 (104, Biolegend,
109830, AB_1186103, 1:400 dilution), BV785-CD3 (145-2C11, Biolegend, 100355, AB_2565969,
1:100 dilution), PE-CD19 (1D3, BD Pharmigen, 553786, AB_395050, 1:400 dilution), Pacific Blue-
Gr1 (RB6-8C5, Biolegend, 108430, AB_893559, 1:200 dilution), BUV395-CD11b (M1/70, BD
Biosciences, 563553, AB_2738276, 1:400 dilution), AF700-MHC-II (I-A/I-E) (M5/114.15.2, E-
Bioscience, 56-5321-82, AB_494009, 1:400 dilution), ef780-CD11c (N418, E-Bioscience, 47-
0114-82, AB_1548652, 1:100 dilution). Cells were fixed using the red blood lysis solution kit (BD
FACS Lysing Solution 10X, 349202) as described in manufacturer’s protocol. Samples were
acquired on a FACSymphony cell analyzer (BD Biosciences). Quantification of total cell numbers
by flow cytometry was done using beads (Beckman Coulter). Samples were acquired on a
Fortessa X20 B (BD Biosciences). Data were analyzed using FlowJo software. CD45.1+ and
CD45.2+ cells were defined as a proportion of single, live granulocytes (CD3-CD19-MHCII-CD11c-
CD11b+GR1+), mononuclear phagocytes (CD3-CD19-MHCII+CD11c+), B cells (CD3-CD19+) or T
cells (CD19-CD3+).

DSS-induced colitis model

Colitis was induced by the addition of 2% DSS (Thermo Scientific, J14489.22) to the
drinking water for 7 days, and were monitored, daily, for weight loss and disease progression.
Moderate severity limits were imposed, with 20% weight loss or two moribund characteristics
judged to be the severity threshold. Mice were euthanized at day 7 post-DSS administration or
earlier if the symptoms of clinical disease (significant weight loss or diarrhea) became apparent.
Colitis progression was assessed using the Disease Activity Index (DAI), as previously described
(75).
Large intestines were fixed in 4% paraformaldehyde and hematoxylin and eosin staining
was performed on 5μm paraffin-embedded cut serial sections (caecum-colon-rectum). Images
were obtained at 20x magnification using Slide Scanner (Zeiss Axio Scan.Z1). Colitis severity
Giampazolias et al 9

was assessed blindly using a histological scoring system as described previously (75). In
addition, spleen mass and colon length were measured to assess the severity of intestinal
inflammation.

Western blot
Dissected tissues from VDRWT and VDRΔIEC were snap frozen in liquid nitrogen and stored at -
80ºC until required. 20-40 mg of each tissue was homogenized in lysis buffer (Cell Lysis Buffer
(10X, Cell Signaling Technology), cOmplete Protease Inhibitor Cocktail (Sigma Aldrich), Halt
Protease & Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific), PMSF Protease Inhibitor
(Thermo Fisher Scientific)) using QIAGEN TissueLyser LT (50 Hz, 10 min). Protein suspensions
were centrifuged twice (1000 x g, 2 min, and 10,000 x g, 10 min) and supernatants harvested.
Protein concentrations were determined using Pierce BCA Protein Assay Kit (Thermo Fisher
Scientific) per the manufacturer’s instructions. Equal quantities of protein from VDRWT and
VDRΔIEC tissues (25 μg small and large intestine, 50 μg kidney, and 100 μg spleen) were denatured
at 75ºC for 10 min in 50 mM DTT and NuPage LDS Loading Dye (4X, Thermo Fisher Scientific).
Samples were run on 7.5 % Mini-PROTEAN TGX Pre-cast Protein Gels in Tris/Glycine/SDS
buffer alongside Precision Plus Protein Dual Color Standards (all Bio-Rad) for 60 min at 100 V.
Protein transfer to 0.2 μm PVDF membrane was performed using the Trans-Blot Turbo Transfer
System (Bio-Rad) per the manufacturer’s instructions. Membrane was blocked in 5 % milk in
Tris-buffered saline with 0.1 % Tween (TBST) for 1 h at room temperature (RT) before addition
of Vitamin D3 Receptor (D2K6W) Rabbit mAb (1:2000, Cell Signaling Technology) in 5 % milk in
TBST overnight at 4ºC. After three washes in TBST, membrane was incubated with GAPDH
(14C10) Rabbit mAb (1:5000, Cell Signaling Technology) in 5 % milk in TBST for 1 h at RT. After
three washes, VDR and GAPDH were detected using secondary anti-rabbit IgG-HRP (7074S)
(1:2000, Cell Signaling Technology) in 5 % milk in TBST for 1 h at RT. Membrane was incubated
with SuperSignal West Pico PLUS chemiluminescent signal (Thermo Fisher Scientific) and
imaged using an Amersham ImageQuant 800.

Quantification of Vitamin D3 metabolites in serum using supercritical fluid

chromatography-Mass spectrometry (SFC-MS)
Supercritical fluid chromatography-mass spectrometry (SFC-MS) was used for the measurement
of VitD3 metabolites in mouse serum based on the elution time of isotope-labelled VitD3 internal
standards (supplemental table 3). Serum samples (200μL) were spiked with 20μL of internal
Giampazolias et al 10

standard 125 ng/mL in methanol 25-HydroxyvitaminD3(6,19,19-d3) (Sigma, 705888) and

incubated for 10 min on ice. Acetonitrile (500μL, HPLC grade) was added to samples and
vortexed for 1 min to precipitate protein and dissociate vitamin D metabolites from vitamin D
binding proteins. Samples were centrifuged at 12,700 rpm for 10 min at 4°C and the supernatant
(SN1) was transferred to a clean Eppendorf 1.5 mL tube. The pellet was re-extracted with 500μL
of acetonitrile, vortexed and centrifuged (12,700 rpm, 10 min at 4°C). The supernatant (SN2) was
transferred and combined with SN1. Combined SN extracts were dried under a gentle stream of
nitrogen. Ethyl acetate (200μL) and ultrapure water (100μL) were added to the dry samples,
vortexed for 1 minute, and centrifuged (12,700 rpm, 5 min at 4°C). The upper organic layer (OL1)
was transferred to a new Eppendorf 1.5 mL tube while the aqueous layer was re-extracted by
adding 200μL ethyl acetate. Samples were vortexed, centrifuged (12,700 rpm, 5 min at 4°C) and
the organic layer (OL2) was combined with OL1. The OL fractions were dried under a gentle
stream of nitrogen and reconstituted in 50μL of methanol. Samples were stored at -20 °C
(overnight) or -80°C (short term) until further SFC-MS analysis.
The SFC separation of 1 μL injection volume was performed using a Waters Acquity UPC2
system, coupled to a Waters Xevo TQ-S micro QqQ MS detector (Waters Corporation, USA),
with a CSH fluoro-phenyl column (50 x 2.1 mm, 1.7 μm) set at 50 °C and the ABPR (automated
back pressure regulator) at 1500 psi. The mobile phase was CO2 (A) and methanol (B) flowing at
a 1 mL/min rate. The elution started at 2% B, maintained for 0.95 min, followed by an increased
to 5% at minute 1.08 and maintained until minute 2.67. Methanol was increased to 10%,
maintained from minute 3.65 to 5.1, and finally returned to 2% at minute 5.61, with a re-
equilibration of 59 seconds. Methanol with 0.2% ammonium formate at 0.25 mL/min flow rate
was used as make-up solvent to enhance the ionization process for vitamin D3 metabolites
before entering the MS. Multiple reaction monitoring (MRM) was used for vitamin D3 metabolite
detection, using positive electrospray ionization mode (ES+) with MRM channels, parent and
daughter ions, cone voltage and collision energies shown in Supplemental Table 3. Capillary
voltage was set to 3.5 kV, desolvation flow to 1000 L/Hr and desolvation temperature to 500 °C.
Data were recorded using MassLynx V4.2 and analysed using TargetLynx XS V4.2 software.
The developed method was subject to validation by assessing the following parameters:
selectivity, linearity, limits of detection (LOD) and quantification (LOQ), recovery, accuracy and
precision. Vitamin D depleted charcoal stripped serum (Golden West Biologicals Inc.), certified
for vitamin D UHPLC-MS/MS applications was used as control matrix for validation experiments.
Linearity was evaluated from 2.5 to 1000 ng/mL (final concentration in serum from 0.625 to 250
Giampazolias et al 11

ng/mL). The LOD and LOQ were set using a signal-to-noise ratio of 3 and 10, respectively. Three
controls, at low (10 ng/mL), medium (50 ng/mL) and high (500 ng/mL) concentrations, were used
to assess recovery, accuracy and precision in triplicate on four non-consecutive days. Samples
with poor recovery of the internal standards were excluded from the analysis.

Shallow shotgun metagenomic sequencing of fecal matter

Transnetyx Microbiome kits containing barcoded sample collection tubes were provided
by Transnetyx (Cordova, TN, USA). Mouse fecal samples were placed in individual
tubes containing DNA stabilization buffer to ensure reproducibility, stability, and traceability, and
shipped for DNA extraction, library preparation, and sequencing by Transnetyx (Cordova, TN
USA). DNA extraction was optimized and fully automated using a robust process
for reproducible extraction of inhibitor-free, high molecular weight genomic DNA that captures
the true microbial diversity of stool samples. DNA extraction is performed using the Qiagen
DNeasy 96 PowerSoil Pro QIAcube HT extraction kit and protocol. After DNA extraction and
quality control (QC), genomic DNA was converted into sequencing libraries using a method
optimized for minimal bias. Unique dual indexed (UDI) adapters were used to ensure that reads
and/or organisms are not misassigned. Library preparation is performed using the KAPA
HyperPlus library preparation protocol. After QC, the libraries were sequenced using the Illumina
NovaSeq instrument and the shotgun sequencing method (a depth of 2 million 2x150 bp read
pairs), which enables species and strain level taxonomic resolution. FASTA/Q files are available
upon request.
Analysis of shotgun sequencing data was performed using JAMS_BW package (v1.8.0)
(76) with its Jan 2022 database publicly available in GitHub as described (77, 78). Taxonomy was
expressed as the last known taxa (LKT), or the taxonomically lowest unambiguous classification
determined for each query sequence, using Kraken’s confidence scoring threshold of
5×10−6 (using the --confidence parameter); p = phylum, c = class, o = order, f = family, g = genus,
s = species. Relative abundance for each LKT within each sample was obtained by dividing the
number of base pairs covering all contigs and unassembled reads pertaining to that LKT by the
total number of non-host base pairs sequenced for that sample. Relative abundances were
expressed as PPM. Gene product annotations were obtained using Prokka. Low abundant
features were filtered if they were not >50 PPM (taxonomy) or >5 PPM (gene product) in at least
5% of samples. For taxonomy, information must come from >70% contigs in at least 50% of
samples and taxon genome completeness must be >5% in at least 5% of samples.
Giampazolias et al 12

For each feature obtained from the shotgun metagenomic sequencing, p values were
calculated using the Mann–Whitney–Wilcoxon U test on PPM relative abundance for that feature
in samples for the following pairwise comparisons: WT + VitD3Standard vs. Gc-/- + VitD3Standard
performed for two independent experiments; WT + VitD3High vs. Gc-/- + VitD3High; WT + VitD3Standard
vs. WT + VitD3High. A meta-analysis method was used to determine common features associated
with changes in the tumor growth phenotype. Fold change was calculated using median within
each group and then averaged (mean) across groups when indicated as Combined. Features
that agreed in directionality of differences across individual comparisons were considered for
the meta-analysis. The combined p value (cp) for meta-analysis of across-group comparisons
was calculated using Fishers p value. For each feature type, the cut-offs for the meta-analysis
were as followed: p-value< 0.2, CP< 0.1, FDR<0.15.

RNA sequencing of colon

Mouse colon was harvested and homogenized in Precellys lysing kit (Bertin Instruments,
P0000917-LYSK0-A) and Precellys 24 tissue homogenizer (Bertin Instruments, P000062-
PEVO0-A). RNA was extracted from the homogenized tissue using the RNeasy Mini Kit (QiaGEN,
74106).
Biological replicate libraries were prepared using the polyA KAPA mRNA HyperPrep Kit
and sequenced on Illumina HiSeq 4000 platform, generating ∼53 million 75bp single end reads
per sample. The RSEM package (79) (version 1.3.30) in conjunction with the STAR alignment
algorithm (79) (version 2.5.2a) was used for the mapping and subsequent gene-level counting of
the sequenced reads with respect to Ensembl mouse GRCm.38.89 version transcriptome.
Normalization of raw count data and differential expression analysis was performed with the
DESeq2 package70 (version 1.24.1) within the R programming environment (version 3.6.0).

Vitamin D and cancer risk epidemiological study in Danish population

The aim of the analysis was to estimate hazard ratio of cancer by site given vitamin D
measurement measured at least one year before cancer diagnosis using data from the Danish
health registries covering the period 2008-2017. The described analyses combine data from the
Central Person Registry, the Cancer Registry and the Register of Laboratory Results for
Research (48, 49). The study is based on data from the Danish nationwide registers
(https://sundhedsdatastyrelsen.dk [sundhedsdatastyrelsen.dk]), which does not require ethical
permission. The Danish Act on Processing of Personal Data has been followed.
Giampazolias et al 13

Central Person Registry: Personal identifiers on everybody in Denmark. Data currently

available from 2.4.1968 until 31.12.2018. Cancer Registry: Data on all cancers diagnosed in
Denmark, currently available from 1.1.1943 until 31.12.2017. Register of Laboratory Results for
Research (RLRR). The RLRR covers all regions of Denmark and includes the results of
biochemistry and immunology tests taken at hospitals and general practitioners with data from
2008 and onwards. The Cancer Registry contains data on all cancers diagnosed in Denmark
from the 1940s and onwards. The study population consisted of all individuals living in Denmark
between 2008 and 2017 with at least one serum measurement of 25-OHD (NPU10267 (P-25-
Hydroxy-Vitamin D(D2+D3)) who were living in Denmark in the period from one year before the
first sample and had never had a cancer diagnosis at the time of the first sample. Exposure was
defined based on the first 25-OHD measurement in RLRR. Follow-up started one year after
sample was taken since vitamin D might be measured as part of testing connected to diagnosis.
People who died, emigrated, or had cancer in the first year after the first sample were excluded.
The first vitamin 25-OHD measurement in the Register of Laboratory Results for Research
starting follow-up one year after sample was taken since vitamin D might be measured as part
of testing connected to diagnosis. A total of 1,496,766 individuals had at least one 25-OHD
measurement with a usable, non-missing result during 2008 and 2016 and satisfied the inclusion
criteria. Among these, 38,905 individuals had measurements above 125 nmol/L that are
considered too high and might cause health problems. Because of this restriction individuals
with 25-OHD values above 125 nmol/L are excluded resulting in a final study population of
1,457,861 individuals. Median length of follow-up (starting one year after sample was taken) was
2.9 years (IQR: 1.3-5.1). Characteristics of individuals including sex, age, year of sample
collection and Charlson’s comorbidity index are available upon request.

Bioinformatic analysis of human tissues and cancer patient data

Normalized read counts for VDR expression were downloaded from the Genotype-Tissue
Expression (GTEx) resource Biobank [https://gtexportal.org]. RSEM normalised gene expression
data was downloaded from TCGA [https://gdac.broadinstitute.org/ ] for 21 cancer datasets. All
analysis were run using R version 3.6.1. Healthy samples and replicates samples were removed.
Only one tumor sample per patient was taken forward for downstream analyses. A Vitamin D-
VDR gene signature consisting of 237 genes was generated using primary VDR target genes that
have been identified using ChIP-sequencing in primary human cells following in vitro stimulation
with 1,25-(OH)2D (11, 42–46) (Supplemental Table 2). Where fold change of gene expression was
Giampazolias et al 14

reported, genes were selected based on log2Fold Change>1.5. The human cell types used
include the monocytic THP1 cell line, blood mononuclear cells, fibroblasts, cancer-associated
fibroblasts, hepatic stellate cells, colorectal cancer cells and lymphoblasts. From these 237
genes, 223 were present in the TCGA datasets (Supplemental Table 2) and patients were ranked
using the sum of the 223 gene Vitamin D-VDR signature. Patients in the top and bottom quartiles
were used to draw the survival plots. To assess the effect of age, sex and stage on the survival,
forest plots were generated with the full data using the ‘survminer’ package (https://CRAN.R-
project.org/package=survminer). In the CP1000+ cohort (47) 230 genes (Supplemental Table 2)
were presented and their sum expression was used as the Vitamin D-VDR signature and its
expression was compared across groups that exhibited differential response to immune
checkpoint inhibitors (CPI). Based on the advice from leading oncologists in Memorial Sloan
Kettering Cancer Centre (MSKCC), “Exceptional response” to CPI was defined as complete (CR)
or partial response (PR) with more or equal to 24 months survival, whereas “Rapid Progressor”
was defined as disease progression (PD) with overall survival (OS) of less than 3 months.

Quantification and statistical analysis

Most statistical analyses were performed using GraphPad Prism software (GraphPad).
Statistical tests are listed in the associated figure legends. Statistical significance between two
groups was determined using an unpaired two-tailed Student’s t test with Welch Correction.
Statistical analyses for two or more groups were done by one- or two-way ANOVA followed by
Bonferroni multiple-comparison post hoc correction. One-way ANOVA was used to compare
average means of two or more groups obtained in a single time point. Two-way ANOVA was
used to compare average means of two or more groups over time (tumor growth profile, mouse
weight and disease activity curves) or average median of multiple groups of different genotypes
(colon length, spleen mass, alpha diversity). The Log-rank “Mantel-Cox” (for two group
comparison) and “for trend’ (for multiple group comparison) tests were used to determine
statistical significance for incidence of tumor rejection or overall survival in mice and cancer
patient data from TCGA. In the gene-enrichment analysis using genes were ranked by the Wald’s
test false discovery rate (FDR)-adjusted p were calculated. PERMANOVA was used to compare
taxa diversity between two groups. Differentially expressed genes were defined as those
showing statistically significant differences between genotype and treatment groups (FDR
<0.05). Gene lists ranked by the Wald statistic were used to look for pathway and biological
process enrichment using the Broad’s GSEA software (version 2.1.0) with gene sets from
Giampazolias et al 15

MSigDB (80) (version 6). Frequency of tumour stages were compared between groups using Chi-
squared test. Time from one year after first vitamin D2+D3 sample until first diagnosis of cancer
was analyzed by a Cox regression model using age as the underlying time scale and adjusting
for sex, age at sample and Charlson’s comorbidity index calculated based on the five years
before the sample was taken (Quan 2005). Data are shown as mean±SD or mean±SEM or
median or hazard ratios with 95% confidence intervals as indicated in the figure legends.
Significance was assumed with *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.
Giampazolias et al 16

Supplementary Fig. 1. Gc-/- mice exhibit microbiome-dependent tumor resistance in

absence of intestinal barrier disruption or inflammation. (A-B) Growth profile of 0.2 x 106
Giampazolias et al 17

5555 BrafV600E cancer cells implanted into: (A) Separately housed WT C57BL/6J (n=12) and co-
housed WT (n=12) and Gc-/- (n=12) groups of mice. Data are representative of two independent
experiments. (B) Separately housed groups of WT or Gc-/- mice that received or not
metronidazole (1 g/L) or neomycin (1 g/L) in the drinking water starting from 2 weeks prior to
tumor inoculation. WT (n=10), Gc-/- (n=10), WT + metronidazole (n=9), Gc-/- + metronidazole
(n=10), WT + neomycin (n=11), Gc-/- + neomycin (n=10). (C-D) Separately housed groups of Gc-
/-
(n=8) and Gc+/+ (n=8) littermate control mice inoculated with 0.2 x 106 5555 BrafV600E cancer
cells. (C) Representative H&E images of small intestine and colon samples (scale bar, 250 μm).
(D) Quantification of immune infiltrates and gut-associated lymphoid tissues in histology samples
from (C). (E) WT or Gc-/- mice (n=6 mice per group) fed with VitD3 standard (2 IU/g) or VitD3 high
(10 IU/g) diet for 3.5 weeks were gavaged with FITC-dextran and fluorescence was measured in
plasma 3h later. (F-I) Quantification of (F) total or (G-I) the indicated immune cell populations in
(G) small intestine, (H) caecum and (I) colon of separately housed groups of WT C57BL/6J (n=9)
or Gc-/- (n=9) mice. Data in (A-B) are presented as tumor volume (mm3) + SEM and are from one
experiment. Data in (D) are presented as median of lymphoid tissue score and are from one
experiment. Data in (E) are presented as mean ± SEM from two independent experiments. Data
in (F-I) are presented as mean frequency (left) or numbers of cells (right) from three independent
experiments. Tumor growth profiles (A-B) were compared using Bonferroni-corrected two-way
ANOVA. Groups in (E) were compared using Bonferroni-corrected one-way ANOVA. Groups in
(D, F-I) were compared using two-tailed unpaired t test with Welch’s correction. *p<0.05,
**p<0.01, ****p< 0.0001; ns, not significant.
Giampazolias et al 18

Supplementary Fig. 2. Gc deficiency does not alter sensitivity of mice to DSS-induced

colitis. (A-D) Separately housed WT C57BL/6J mouse groups (n=10 per group) that received FT
from WT C57BL/6J or Gc-/- donors (on days -14 and -12) prior to administration or not of 2%
DSS (day 0) in the drinking water. DSS-induced colitis was measured after 7 days of switching
to DSS water. (A) Representative H&E images of transversely cut colon segments (scale bar, 250
μm). (B) Histopathological scoring of colitis severity. (C) Quantification of normalised colon mass
(left) and disease activity index (right) measurement over time following DSS administration. (D)
Quantification of colon length (cm) (left) and spleen mass (mg) (right) over time following DSS
administration. Data in (B-C) are presented as mean ± SEM, data in (D) are presented as median
and are from one experiment. Data in (B-D) were compared using Bonferroni-corrected two-way
ANOVA. *p<0.05, **p<0.01, ****p< 0.0001; ns, not significant.
Giampazolias et al 19

Supplementary Fig. 3. Fecal transplants from Gc-/- mice augment tumor control dependent
on CD8+ T cells and MyD88. (A-B) Growth profile of 0.2 x 106 5555 BrafV600E cancer cells
implanted into the indicated separately housed groups of mice that received (on days -14 and -
12 prior to tumor inoculation) FT from WT C57BL/6J or Gc-/- donors. (A) WT + WT FT (n=10), WT
+ Gc-/- FT (n=10), Tap1-/- + WT FT (n=10), Tap1-/- + Gc-/- FT (n=10). (B) WT + WT FT (n=9), WT +
Gc-/- FT (n=10), Cgas-/- + WT FT (n=11), Cgas-/- + Gc-/- FT (n=11), Trif-/- + WT FT (n=10), Trif-/- +
Gc-/- FT (n=11), Myd88-/- + WT FT (n=9), Myd88-/- + Gc-/- FT (n=10). (C) Quantification of the
indicated immune populations in the blood of irradiated CD45.1 WT mice whose bone marrow
(BM) was reconstituted using BM from CD45.2 WT or Myd88-/- donors. WT (WT BM) + WT FT
(n=11), WT (WT BM) + Gc-/- FT (n=12), WT (Myd88-/- BM) + WT FT (n=10), WT (Myd88-/- BM) Gc-/-
FT (n=10). Data in (A-B) are presented as tumor volume (mm3) + SEM. Data in (A) are
representative of two independent experiments. Data in (B) are from one experiment apart from
the growth profiles of WT and Myd88-/- mice that are representative of two independent
experiments. Data in (C) are presented as mean frequency ± SEM and are representative of two
experiments. Data in (A-C) were compared using Bonferroni-corrected two-way ANOVA. ***p<
0.001, ****p< 0.0001; ns, not significant.
Giampazolias et al 20

Supplementary Fig. 4. Mice with increased VitD dietary supplementation develop

microbiome-dependent transmissible tumor resistance. (A) Quantification of VitD3
metabolites in the serum of separately housed groups of WT C57BL/6J (n=5 per group) or Gc-/-
(n=5 per group) mice that were fed with VitD3 standard (2 IU/g), deficient (0 IU/g) or high (10 IU/g)
diet starting from 3.5 weeks. (B-E) Growth profile of 0.2 x 106 5555 BrafV600E cancer cells
implanted into: B) Separately housed groups of C57BL/6J germ-free (GF) mice (n=10 per group)
that were fed with VitD3 deficient (2 IU/g), or high (10 IU/g) diet starting from 3.5 weeks prior to
tumor inoculation. (C) Separately housed groups of C57BL/6J WT mice (n=10 per group) that
were fed with VitD3 standard (2 IU/g), or high (10 IU/g) diet starting from 3.5 weeks prior to tumor
inoculation and received or not vancomycin (0.5 g/L) in the drinking water starting from 2 weeks
prior to tumor inoculation. (D) C57BL/6J WT mice (n=10 per group) that received FT (on day -14
and -12) from donors treated as in (C). (E) C57BL/6 mice were purchased from two different
commercial vendors [C57BL/6JOlaHsd2 Envigo (n=10 per group], [C57BL/6J Charles River
(n=10 vs. n=9)] or were procured from the Crick [C57BL/6J Crick (n=8 vs. n=8)] and separately
Giampazolias et al 21

housed at CRUK Manchester Institute (CRUK MI). All groups received FT (on day -14 and -12)
from C57BL/6J Crick donors that were fed with VitD3 standard (2 IU/g) or high (10 IU/g) diet.
Data in (A) are presented as mean concentration ± SEM and are from one experiment. Data in
(B-E) are presented mean tumor volume (mm3) + SEM and are from one experiment. Data in (A-
E) were compared using Bonferroni-corrected two-way ANOVA. *p<0.05, **p<0.01, ****p<
0.0001; ns, not significant.
Giampazolias et al 22

Supplementary Fig. 5. RNA sequencing of the colon of WT mice fed with standard or high
VitD diet. (A-D) RNA sequencing of colons from separately housed groups of WT C57BL/6J
mice (n=5 per group) that were fed with VitD3 standard (2 IU/g) or high (10 IU/g) diet for 3.5
weeks. (A) Gene expression profile (heatmap) of individual colon samples. (B) Gene ontology
Giampazolias et al 23

biological process (GOBP) analysis in the high (VitD3High diet) comparing to the (VitD3Low diet)
group of colon samples mapped to Gene Ontology Biological Process (GOBP) database. (C-D)
Proportion of CIBERSORT-gene signatures of (C) myeloid and (D) lymphoid immune populations
within individual colon samples. Gene expression in (A) is presented as normalized read counts
across samples for each gene (z-scores). In (B) all genes were ranked by the Wald’s test and
FDR-adjusted p values (q values) were calculated. In (A-B), data are normalized gene expression
in (A) and normalized enrichment score in (B) with positive values in red, negative values in blue.
In (C-D) data are presented as frequency mean ± SEM of gene signatures within a sample. In (C-
D) frequency means of gene signatures between groups were compared using Bonferroni-
corrected two-way ANOVA. ****p< 0.0001; ns, not significant. Data in (A-D) are from one
experiment.
Giampazolias et al 24

Supplementary Fig. 6. Bacterial taxonomic diversity in the fecal matter of mice with high
or low VitD availability. (A-D) Sequenced bacterial taxa diversity (A, C) within (alpha diversity,
inverse Simpson index) and (B, D) across (beta diversity, principal component analysis) fecal
samples from separately housed groups of WT C57BL/6J or Gc-/- mice that were fed with VitD3
standard (2 IU/g) or deficient (0 IU/g) (A, B) or VitD3 standard (2 IU/g) or high (10 IU/g) (C, D) diet
Giampazolias et al 25

for 3.5 weeks. (A, B) WT + VitD3Standard (n=10), Gc-/- + VitD3Standard (n=9), WT + VitD3Deficient (n=10),
Gc-/- + VitD3Deficient (n=10). (C-D) WT + VitD3Standard (n=12), Gc-/- + VitD3Standard (n=11), WT + VitD3High
(n=13), Gc-/- + VitD3High (n=13). Data in (A, C) are presented as inverse Simpson index box-and-
whisker plots with median, 25th and 75th percentiles represented by the box and min/max by
the whiskers and are from one experiment. Data in (B, D) are presented as projected data of
variances of two individual principal components (PC1 and PC2). Inverse Simpson indexes (A,
C) were compared using Bonferroni-corrected two-way ANOVA. Taxa composition between
groups in (B, D) were compared using PERMANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<
0.0001; ns, not significant.
Giampazolias et al 26

Supplementary Fig. 7. Phylum abundance in the fecal matter of mice with high or low VitD
availability. (A-B) Relative abundance bar plots of bacterial phyla in fecal samples from
separately housed groups of WT C57BL/6J or Gc-/- mice that were fed with VitD3 standard (2
IU/g) or deficient (0 IU/g) (A) or VitD3 standard (2 IU/g) or high (10 IU/g) (B) diet for 3.5 weeks. (A)
WT + VitD3Standard (n=10), Gc-/- + VitD3Standard (n=9), WT + VitD3Deficient (n=10), Gc-/- + VitD3Deficient (n=10).
(B) WT + VitD3Standard (n=12), Gc-/- + VitD3Standard (n=11), WT + VitD3High (n=13), Gc-/- + VitD3High (n=13).
Data in (A, B) are shown as stacked bar plots for individual mice separated by groups (left) and
averaged per group (center), and as relative abundance (PPM) box-and-whisker plots with
median, 25th and 75th percentiles represented by the box and min/max by the whiskers (right).
Giampazolias et al 27

Taxa abundances (A, B, right panel) were compared using Bonferroni-corrected two-way
ANOVA. *p<0.05, **p< 0.01; ns, not significant.
Giampazolias et al 28

Supplementary Fig. 8. Order abundance in the fecal matter of mice with high or low VitD
availability. (A-B) Relative abundance bar plots of bacterial orders in fecal samples from
separately housed groups of WT C57BL/6J or Gc-/- mice that were fed with VitD3 standard (2
IU/g) or deficient (0 IU/g) (A) or VitD3 standard (2 IU/g) or high (10 IU/g) (B) diet for 3.5 weeks. (A)
Giampazolias et al 29

WT + VitD3Standard (n=10), Gc-/- + VitD3Standard (n=9), WT + VitD3Deficient (n=10), Gc-/- + VitD3Deficient (n=10).
(B) WT + VitD3Standard (n=12), Gc-/- + VitD3Standard (n=11), WT + VitD3High (n=13), Gc-/- + VitD3High (n=13).
Data in (A, B) are shown as stacked bar plots for individual mice separated by groups (left) and
averaged per group (center), and as relative abundance (PPM) box-and-whisker plots with
median, 25th and 75th percentiles represented by the box and min/max by the whiskers (right).
Taxa abundances (A, B, right panel) were compared using Bonferroni-corrected two-way
ANOVA. *p<0.05, **p< 0.01, ****p< 0.0001; ns, not significant.
Giampazolias et al 30

Supplementary Fig. 9. Meta analyses of taxonomic and functional bacterial features in the
fecal matter of mice with high or low VitD availability. (A-C) Meta-analyses of indicated
comparisons for (A) bacterial gene products and (B, C) taxonomy associated with differences in
Giampazolias et al 31

VitD availability: Meta1 (left), Exp 1 WT + VitD3Standard (n=10) vs. Gc-/- + VitD3Standard (n=9) and Exp2
WT + VitD3Standard (n=12) vs. Gc-/- + VitD3Standard (n=11); Meta2 (center), WT + VitD3Standard (n=12) vs.
Gc-/- + VitD3Standard (n=11) and WT + VitD3High (n=13) vs. Gc-/- + VitD3High (n=13); Meta3 (right), WT +
VitD3Standard (n=12) vs. Gc-/- + VitD3Standard (n=11) and WT + VitD3High (n=13) vs. WT + VitD3Standard
(n=12). Only features that showed consistent direction of fold change across both comparisons
were used for the meta-analyses and plotted. Data in (A, B) are presented as log2 median fold
change for each comparison and in (C) as average of log2 median fold change across both
comparisons in each meta-analysis. In (A, B and C), p values were calculated using the Mann–
Whitney–Wilcoxon U test on relative abundance (PPM) for that feature in samples within each
group for pairwise comparisons. The combined p value (cp) for meta-analysis of within-group
comparisons was calculated using Fishers P value. In (A, B) the number of significant/total
features in each analysis are shown above each plot. In (C), significantly different taxonomy
common across all three analyses are highlighted. For each feature type, the cut-offs for the
meta-analysis were: p< 0.2, cp< 0.1, FDR<0.15 (red); not significant (grey).
Giampazolias et al 32

Supplementary Fig. 10. B. fragilis is associated with VitD3 availability and enhanced tumor
resistance. (A-C) Sequence analysis of bacterial taxa in fecal samples from separately housed
Giampazolias et al 33

groups of WT C57BL/6J or Gc-/- mice that were fed with VitD3 standard (2 IU/g) or deficient (0
IU/g) or VitD3 standard (2 IU/g) or high (10 IU/g) diet for 3.5 weeks. Relative abundance of (A)
Bacteroides fragilis and (B) Prevotella brevis. (C) Differentially abundant taxa between WT and
Gc-/- present when mice are given VitD3Standard but not VitD3Deficient; (left) Venn diagram showing
number of overlapping and exclusive differentially abundant taxa across both diets and (right)
volcano plot highlighting 7 taxa exclusive to VitD3Standard. (A, B, and C) WT + VitD3Standard (n=10),
Gc-/- + VitD3Standard (n=9), WT + VitD3Deficient (n=10), Gc-/- + VitD3Deficient (n=10). (A, B) WT + VitD3Standard
(n=12), Gc-/- + VitD3Standard (n=11), WT + VitD3High (n=13), Gc-/- + VitD3High (n=13). (D) Growth profile
of 0.2 x 106 5555 BrafV600E cancer cells implanted into separately housed groups of
C57BL/6JOlaHsd2 Envigo (n=10 per group), C57BL/6J Charles River (n=9 vs. n=10), C57BL/6J
Crick (n=10 vs. n=9) before receiving B. fragilis or vehicle at CRUK Manchester Institute (CRUK
MI). Mice received 109 B. fragilis by oral gavage on days -14, -12 and -10 prior to tumor
inoculation. (E) Q-PCR quantification of 16S rRNA gene of B. fragilis in fecal matter of mice that
were fed VitD3 standard or deficient diet and received B.fragilis or vehicle, relative to abundance
of total 16S rRNA gene. WT-VitD3Standard + vehicle (n=17), WT-VitD3Deficient+ vehicle (n=18), WT-
VitD3Standard + B.fragilis (n=16), WT-VitD3Deficient + B.fragilis (n=19). (F) Growth profile of 0.2 x 106
5555 BrafV600E cancer cells implanted into separately housed groups of WT C57BL/6J mice (n=10
per group) fed with VitD3 standard or VitD3 deficient diet, starting 3.5 weeks before receiving P.
brevis or vehicle. Mice received 109 P. brevis by oral gavage on days -14, -12 and -10 prior to
tumor inoculation. Data in (A, B) are presented as relative abundance (PPM) box-and-whisker
plots with median, 25th and 75th percentiles represented by the box and min/max by the
whiskers and are from one experiment. Data in (C) is presented as log2 median fold change vs.
-log(FDR). Data in (D) are presented as mean ± SD of two independent experiments. Data in (D
and F) are presented as tumor volume (mm3) + SEM of one experiment (D) or are representative
of two independent experiments (F). Data in (C) were compared using the Mann–Whitney–
Wilcoxon U test on PPM in samples within each diet for pairwise comparisons with significance
indicated by p < 0.05, FDR < 0.15 within each comparison. Data in (A, B, D, E, and F) were
compared using Bonferroni-corrected two-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<
0.0001; ns, not significant.
Giampazolias et al 34

Supplementary Fig. 11. Genome-wide association studies and association of patient

survival, tumor stage and response to checkpoint inhibitors with gene expression in
human tumors. (A) Schematic representation of the correlation of VitD3 metabolites and proteins
involved in VitD3 biosynthesis, availability or biological function with cancer risk (red square),
Giampazolias et al 35

microbiota alterations (purple square) or immune parameters in health and disease (green
square). The asterisk indicates polymorphisms in the indicated genes that are identified in
genome-wide association studies correlated with at least one of the four parameters listed
(coloured squares). (B) Human tissue expression of VDR from the Genotype-Tissue Expression
database. (C) Prognostic value of VDR expression for overall survival and hazard ratio comparing
samples with the lowest (VDRLow) versus highest (VDRHigh) expression in the indicated TCGA
datasets. Skin cutaneous melanoma (SKMC, n=460), sarcoma (SARC, n=259), liver
hepatocellular carcinoma (LIHC, n=370) bottom and top 25% of patient cohort. (D) Prognostic
value of VDR expression for tumor stage comparing samples with the lowest (VDRLow) versus
highest (VDRHigh) expression in the indicated TCGA datasets. Breast cancer (BRCA, n=1092),
prostate adenocarcinoma (PRAD, n=497) bottom and top 25% of patient cohort. (E) Expression
of the indicated immune related gene signatures and genes comparing samples with the (VitD-
VDR gene signHigh) versus the lowest (VitD-VDR gene signLow) expression in the indicated TCGA
datasets. (F) VDR expression in samples with no versus exceptional response (top) and rapid
versus standard disease progression (bottom) of patients (n=1008) treated with checkpoint
inhibitors (CPI1000+ cohort). In (C, D) data are presented as mean of log2 normalized expression
± SEM. In (E) data are presented as log2 fold change (FC) of gene or gene signature expression
comparing VitD-VDR gene sign High/Low groups (colour intensity represent the magnitude of
FC and the size of the circle depicts the magnitude of the probability value). In (F) data are
presented as log2 normalized expression box-and-whisker plot with median, 25th and 75th
percentiles represented by the box and min/max by the whiskers. Survival (Kaplan-Meier) curves
in (C) were compared using Log-rank (Mantel-Cox) test. In (C) hazard ratios (HR) with 95%
confidence interval showed. In (C, D and E) expression of genes and gene signatures between
groups were compared using two-tailed unpaired t test with Welch’s correction. In (D) frequency
of tumour stage was compared between groups using Chi-squared test. In (F) gene expression
between the groups was compared using Wilcoxon signed-rank test. *p<0.05, **p<0.01, ***p<
0.001; ns, not significant.
Giampazolias et al 36
Giampazolias et al 37

Supplementary Fig. 12. Schematic summary of the findings. (A-B) Mice with increased VitD
activity display (A) endogenous and (B) transmissible tumor resistance. (C) VitD acts on intestinal
epithelial cells, which modulate the gut microbiome increasing a niche for B. fragilis, which
boosts T cell responses to cancer. Limited VitD activity due to dietary restriction of VitD and
sequestration by plasma Gc globulin dampens microbiome-dependent anti-cancer immunity.

Supplemental Table 1. Polymorphisms in genes regulating VitD availability correlate with

cancer risk, alterations in microbiota and/or changes in immune parameters.

Supplemental Table 2. VitD-VDR gene signature in TCGA and CP1000+ cohorts.

Supplemental Table 3. Elution time of isotope-labelled VitD3 internal standards in SFC-MS

analysis.
References and Notes
1. E. A. Yamamoto, T. N. Jørgensen, Relationships Between Vitamin D, Gut Microbiome, and
Systemic Autoimmunity. Front. Immunol. 10, 3141 (2020).
doi:10.3389/fimmu.2019.03141 Medline
2. M. Medrano, E. Carrillo-Cruz, I. Montero, J. A. Perez-Simon, Vitamin D: Effect on
Haematopoiesis and Immune System and Clinical Applications. Int. J. Mol. Sci. 19, 2663
(2018). doi:10.3390/ijms19092663 Medline
3. R. M. Lucas, S. Gorman, S. Geldenhuys, P. H. Hart, Vitamin D and immunity. F1000Prime
Rep. 6, 118 (2014). doi:10.12703/P6-118 Medline
4. A. Clark, N. Mach, Role of Vitamin D in the Hygiene Hypothesis: The Interplay between
Vitamin D, Vitamin D Receptors, Gut Microbiota, and Immune Response. Front.
Immunol. 7, 627 (2016). doi:10.3389/fimmu.2016.00627 Medline
5. M. Waterhouse, B. Hope, L. Krause, M. Morrison, M. M. Protani, M. Zakrzewski, R. E.
Neale, Vitamin D and the gut microbiome: A systematic review of in vivo studies. Eur. J.
Nutr. 58, 2895–2910 (2019). doi:10.1007/s00394-018-1842-7 Medline
6. J. Wang, L. B. Thingholm, J. Skiecevičienė, P. Rausch, M. Kummen, J. R. Hov, F.
Degenhardt, F.-A. Heinsen, M. C. Rühlemann, S. Szymczak, K. Holm, T. Esko, J. Sun,
M. Pricop-Jeckstadt, S. Al-Dury, P. Bohov, J. Bethune, F. Sommer, D. Ellinghaus, R. K.
Berge, M. Hübenthal, M. Koch, K. Schwarz, G. Rimbach, P. Hübbe, W.-H. Pan, R.
Sheibani-Tezerji, R. Häsler, P. Rosenstiel, M. D’Amato, K. Cloppenborg-Schmidt, S.
Künzel, M. Laudes, H.-U. Marschall, W. Lieb, U. Nöthlings, T. H. Karlsen, J. F. Baines,
A. Franke, Genome-wide association analysis identifies variation in vitamin D receptor
and other host factors influencing the gut microbiota. Nat. Genet. 48, 1396–1406 (2016).
doi:10.1038/ng.3695 Medline
7. K. K. Deeb, D. L. Trump, C. S. Johnson, Vitamin D signalling pathways in cancer: Potential
for anticancer therapeutics. Nat. Rev. Cancer 7, 684–700 (2007). doi:10.1038/nrc2196
Medline
8. S.-M. Jeon, E.-A. Shin, Exploring vitamin D metabolism and function in cancer. Exp. Mol.
Med. 50, 1–14 (2018). doi:10.1038/s12276-018-0038-9 Medline
9. C. Carlberg, A. Muñoz, An update on vitamin D signaling and cancer. Semin. Cancer Biol. 79,
217–230 (2022). doi:10.1016/j.semcancer.2020.05.018 Medline
10. M. H. Sherman, R. T. Yu, D. D. Engle, N. Ding, A. R. Atkins, H. Tiriac, E. A. Collisson, F.
Connor, T. Van Dyke, S. Kozlov, P. Martin, T. W. Tseng, D. W. Dawson, T. R.
Donahue, A. Masamune, T. Shimosegawa, M. V. Apte, J. S. Wilson, B. Ng, S. L. Lau, J.
E. Gunton, G. M. Wahl, T. Hunter, J. A. Drebin, P. J. O’Dwyer, C. Liddle, D. A.
Tuveson, M. Downes, R. M. Evans, Vitamin D receptor-mediated stromal
reprogramming suppresses pancreatitis and enhances pancreatic cancer therapy. Cell 159,
80–93 (2014). doi:10.1016/j.cell.2014.08.007 Medline
11. G. Ferrer-Mayorga, G. Gómez-López, A. Barbáchano, A. Fernández-Barral, C. Peña, D. G.
Pisano, R. Cantero, F. Rojo, A. Muñoz, M. J. Larriba, Vitamin D receptor expression and
 associated gene signature in tumour stromal fibroblasts predict clinical outcome in
colorectal cancer. Gut 66, 1449–1462 (2017). doi:10.1136/gutjnl-2015-310977 Medline
12. J. Wactawski-Wende, J. M. Kotchen, G. L. Anderson, A. R. Assaf, R. L. Brunner, M. J.
O’Sullivan, K. L. Margolis, J. K. Ockene, L. Phillips, L. Pottern, R. L. Prentice, J.
Robbins, T. E. Rohan, G. E. Sarto, S. Sharma, M. L. Stefanick, L. Van Horn, R. B.
Wallace, E. Whitlock, T. Bassford, S. A. A. Beresford, H. R. Black, D. E. Bonds, R. G.
Brzyski, B. Caan, R. T. Chlebowski, B. Cochrane, C. Garland, M. Gass, J. Hays, G.
Heiss, S. L. Hendrix, B. V. Howard, J. Hsia, F. A. Hubbell, R. D. Jackson, K. C. Johnson,
H. Judd, C. L. Kooperberg, L. H. Kuller, A. Z. LaCroix, D. S. Lane, R. D. Langer, N. L.
Lasser, C. E. Lewis, M. C. Limacher, J. E. Manson, Women’s Health Initiative
Investigators, Calcium plus vitamin D supplementation and the risk of colorectal cancer.
N. Engl. J. Med. 354, 684–696 (2006). doi:10.1056/NEJMoa055222 Medline
13. M. Jenab, H. B. Bueno-de-Mesquita, P. Ferrari, F. J. B. van Duijnhoven, T. Norat, T.
Pischon, E. H. J. M. Jansen, N. Slimani, G. Byrnes, S. Rinaldi, A. Tjønneland, A. Olsen,
K. Overvad, M.-C. Boutron-Ruault, F. Clavel-Chapelon, S. Morois, R. Kaaks, J.
Linseisen, H. Boeing, M. M. Bergmann, A. Trichopoulou, G. Misirli, D. Trichopoulos, F.
Berrino, P. Vineis, S. Panico, D. Palli, R. Tumino, M. M. Ros, C. H. van Gils, P. H.
Peeters, M. Brustad, E. Lund, M.-J. Tormo, E. Ardanaz, L. Rodríguez, M.-J. Sánchez, M.
Dorronsoro, C. A. Gonzalez, G. Hallmans, R. Palmqvist, A. Roddam, T. J. Key, K.-T.
Khaw, P. Autier, P. Hainaut, E. Riboli, Association between pre-diagnostic circulating
vitamin D concentration and risk of colorectal cancer in European populations:a nested
case-control study. BMJ 340, b5500 (2010). doi:10.1136/bmj.b5500 Medline
14. C. G. Woolcott, L. R. Wilkens, A. M. Y. Nomura, R. L. Horst, M. T. Goodman, S. P.
Murphy, B. E. Henderson, L. N. Kolonel, L. Le Marchand, Plasma 25-hydroxyvitamin D
levels and the risk of colorectal cancer: The multiethnic cohort study. Cancer Epidemiol.
Biomarkers Prev. 19, 130–134 (2010). doi:10.1158/1055-9965.EPI-09-0475 Medline
15. S. Gandini, M. Boniol, J. Haukka, G. Byrnes, B. Cox, M. J. Sneyd, P. Mullie, P. Autier,
Meta-analysis of observational studies of serum 25-hydroxyvitamin D levels and
colorectal, breast and prostate cancer and colorectal adenoma. Int. J. Cancer 128, 1414–
1424 (2011). doi:10.1002/ijc.25439 Medline
16. Y. Ma, P. Zhang, F. Wang, J. Yang, Z. Liu, H. Qin, Association between vitamin D and risk
of colorectal cancer: A systematic review of prospective studies. J. Clin. Oncol. 29,
3775–3782 (2011). doi:10.1200/JCO.2011.35.7566 Medline
17. J. De Smedt, S. Van Kelst, V. Boecxstaens, M. Stas, K. Bogaerts, D. Vanderschueren, C.
Aura, K. Vandenberghe, D. Lambrechts, P. Wolter, O. Bechter, A. Nikkels, T. Strobbe,
G. Emri, V. Marasigan, M. Garmyn, Vitamin D supplementation in cutaneous malignant
melanoma outcome (ViDMe): A randomized controlled trial. BMC Cancer 17, 562
(2017). doi:10.1186/s12885-017-3538-4 Medline
18. L. Yang, H. Chen, M. Zhao, P. Peng, Prognostic value of circulating vitamin D binding
protein, total, free and bioavailable 25-hydroxy vitamin D in patients with colorectal
cancer. Oncotarget 8, 40214–40221 (2017). doi:10.18632/oncotarget.16597 Medline
19. J. E. Manson, N. R. Cook, I.-M. Lee, W. Christen, S. S. Bassuk, S. Mora, H. Gibson, D.
Gordon, T. Copeland, D. D’Agostino, G. Friedenberg, C. Ridge, V. Bubes, E. L.
 Giovannucci, W. C. Willett, J. E. Buring, VITAL Research Group, Vitamin D
Supplements and Prevention of Cancer and Cardiovascular Disease. N. Engl. J. Med. 380,
33–44 (2019). doi:10.1056/NEJMoa1809944 Medline
20. K. Ng, H. S. Nimeiri, N. J. McCleary, T. A. Abrams, M. B. Yurgelun, J. M. Cleary, D. A.
Rubinson, D. Schrag, R. Miksad, A. J. Bullock, J. Allen, D. Zuckerman, E. Chan, J. A.
Chan, B. M. Wolpin, M. Constantine, D. J. Weckstein, M. A. Faggen, C. A. Thomas, C.
Kournioti, C. Yuan, C. Ganser, B. Wilkinson, C. Mackintosh, H. Zheng, B. W. Hollis, J.
A. Meyerhardt, C. S. Fuchs, Effect of High-Dose vs Standard-Dose Vitamin D3
Supplementation on Progression-Free Survival Among Patients With Advanced or
Metastatic Colorectal Cancer: The SUNSHINE Randomized Clinical Trial. JAMA 321,
1370–1379 (2019). doi:10.1001/jama.2019.2402 Medline
21. M. Urashima, H. Ohdaira, T. Akutsu, S. Okada, M. Yoshida, M. Kitajima, Y. Suzuki, Effect
of Vitamin D Supplementation on Relapse-Free Survival Among Patients With Digestive
Tract Cancers: The AMATERASU Randomized Clinical Trial. JAMA 321, 1361–1369
(2019). doi:10.1001/jama.2019.2210 Medline
22. D. D. Bikle, Vitamin D metabolism, mechanism of action, and clinical applications. Chem.
Biol. 21, 319–329 (2014). doi:10.1016/j.chembiol.2013.12.016 Medline
23. J. G. Haddad, Y. Z. Hu, M. A. Kowalski, C. Laramore, K. Ray, P. Robzyk, N. E. Cooke,
Identification of the sterol- and actin-binding domains of plasma vitamin D binding
protein (Gc-globulin). Biochemistry 31, 7174–7181 (1992). doi:10.1021/bi00146a021
Medline
24. F. F. Safadi, P. Thornton, H. Magiera, B. W. Hollis, M. Gentile, J. G. Haddad, S. A.
Liebhaber, N. E. Cooke, Osteopathy and resistance to vitamin D toxicity in mice null for
vitamin D binding protein. J. Clin. Invest. 103, 239–251 (1999). doi:10.1172/JCI5244
Medline
25. E. G. Duchow, N. E. Cooke, J. Seeman, L. A. Plum, H. F. DeLuca, Vitamin D binding
protein is required to utilize skin-generated vitamin D. Proc. Natl. Acad. Sci. U.S.A. 116,
24527–24532 (2019). doi:10.1073/pnas.1915442116 Medline
26. C. M. Henderson, S. L. Fink, H. Bassyouni, B. Argiropoulos, L. Brown, T. J. Laha, K. J.
Jackson, R. Lewkonia, P. Ferreira, A. N. Hoofnagle, J. L. Marcadier, Vitamin D–Binding
Protein Deficiency and Homozygous Deletion of the GC Gene. N. Engl. J. Med. 380,
1150–1157 (2019). doi:10.1056/NEJMoa1807841 Medline
27. J. P. Böttcher, C. Reis e Sousa, The Role of Type 1 Conventional Dendritic Cells in Cancer
Immunity. Trends Cancer 4, 784–792 (2018). doi:10.1016/j.trecan.2018.09.001 Medline
28. S. K. Wculek, F. J. Cueto, A. M. Mujal, I. Melero, M. F. Krummel, D. Sancho, Dendritic
cells in cancer immunology and immunotherapy. Nat. Rev. Immunol. 20, 7–24 (2020).
doi:10.1038/s41577-019-0210-z Medline
29. J. Canton, H. Blees, C. M. Henry, M. D. Buck, O. Schulz, N. C. Rogers, E. Childs, S.
Zelenay, H. Rhys, M.-C. Domart, L. Collinson, A. Alloatti, C. J. Ellison, S. Amigorena,
V. Papayannopoulos, D. C. Thomas, F. Randow, C. Reis e Sousa, The receptor DNGR-1
signals for phagosomal rupture to promote cross-presentation of dead-cell-associated
antigens. Nat. Immunol. 22, 140–153 (2021). doi:10.1038/s41590-020-00824-x Medline
30. C. M. Henry, C. A. Castellanos, C. Reis e Sousa, DNGR-1-mediated cross-presentation of
dead cell-associated antigens. Semin. Immunol. 66, 101726 (2023).
doi:10.1016/j.smim.2023.101726 Medline
31. E. Giampazolias, O. Schulz, K. H. J. Lim, N. C. Rogers, P. Chakravarty, N. Srinivasan, O.
Gordon, A. Cardoso, M. D. Buck, E. Z. Poirier, J. Canton, S. Zelenay, S. Sammicheli, N.
Moncaut, S. Varsani-Brown, I. Rosewell, C. Reis e Sousa, Secreted gelsolin inhibits
DNGR-1-dependent cross-presentation and cancer immunity. Cell 184, 4016–4031.e22
(2021). doi:10.1016/j.cell.2021.05.021 Medline
32. K. H. J. Lim, E. Giampazolias, O. Schulz, N. C. Rogers, A. Wilkins, E. Sahai, J. Strid, C.
Reis e Sousa, Loss of secreted gelsolin enhances response to anticancer therapies. J.
Immunother. Cancer 10, e005245 (2022). doi:10.1136/jitc-2022-005245 Medline
33. W. M. Lee, R. M. Galbraith, The extracellular actin-scavenger system and actin toxicity. N.
Engl. J. Med. 326, 1335–1341 (1992). doi:10.1056/NEJM199205143262006 Medline
34. O. Koivisto, A. Hanel, C. Carlberg, Key Vitamin D Target Genes with Functions in the
Immune System. Nutrients 12, 1140 (2020). doi:10.3390/nu12041140 Medline
35. M. Amling, M. Priemel, T. Holzmann, K. Chapin, J. M. Rueger, R. Baron, M. B. Demay,
Rescue of the skeletal phenotype of vitamin D receptor-ablated mice in the setting of
normal mineral ion homeostasis: Formal histomorphometric and biomechanical analyses.
Endocrinology 140, 4982–4987 (1999). doi:10.1210/endo.140.11.7110 Medline
36. A. M. Mondul, I. M. Shui, K. Yu, R. C. Travis, V. L. Stevens, D. Campa, F. R. Schumacher,
R. G. Ziegler, H. B. Bueno-de-Mesquita, S. Berndt, E. D. Crawford, S. M. Gapstur, J. M.
Gaziano, E. Giovannucci, C. A. Haiman, B. E. Henderson, D. J. Hunter, M. Johansson, T.
J. Key, L. Le Marchand, S. Lindström, M. L. McCullough, C. Navarro, K. Overvad, D.
Palli, M. Purdue, M. J. Stampfer, S. J. Weinstein, W. C. Willett, M. Yeager, S. J.
Chanock, D. Trichopoulos, L. N. Kolonel, P. Kraft, D. Albanes, Genetic variation in the
vitamin D pathway in relation to risk of prostate cancer—Results from the breast and
prostate cancer cohort consortium. Cancer Epidemiol. Biomarkers Prev. 22, 688–696
(2013). doi:10.1158/1055-9965.EPI-13-0007-T Medline
37. L. Zhou, X. Zhang, X. Chen, L. Liu, C. Lu, X. Tang, J. Shi, M. Li, M. Zhou, Z. Zhang, L.
Xiao, M. Yang, GC Glu416Asp and Thr420Lys polymorphisms contribute to
gastrointestinal cancer susceptibility in a Chinese population. Int. J. Clin. Exp. Med. 5,
72–79 (2012). Medline
38. S. Karami, G. Andreotti, S. Koutros, K. H. Barry, L. E. Moore, S. Han, J. A. Hoppin, D. P.
Sandler, J. H. Lubin, L. A. Burdette, J. Yuenger, M. Yeager, L. E. B. Freeman, A. Blair,
M. C. R. Alavanja, Pesticide exposure and inherited variants in vitamin D pathway genes
in relation to prostate cancer. Cancer Epidemiol. Biomarkers Prev. 22, 1557–1566
(2013). doi:10.1158/1055-9965.EPI-12-1454 Medline
39. M. Peña-Chilet, M. Ibarrola-Villava, M. Martin-González, M. Feito, C. Gomez-Fernandez,
D. Planelles, G. Carretero, A. Lluch, E. Nagore, G. Ribas, rs12512631 on the group
specific complement (vitamin D-binding protein GC) implicated in melanoma
susceptibility. PLOS ONE 8, e59607 (2013). doi:10.1371/journal.pone.0059607 Medline
40. L. N. Anderson, M. Cotterchio, D. E. C. Cole, J. A. Knight, Vitamin D-related genetic
variants, interactions with vitamin D exposure, and breast cancer risk among Caucasian
women in Ontario. Cancer Epidemiol. Biomarkers Prev. 20, 1708–1717 (2011).
doi:10.1158/1055-9965.EPI-11-0300 Medline
41. R. F. Chun, New perspectives on the vitamin D binding protein. Cell Biochem. Funct. 30,
445–456 (2012). doi:10.1002/cbf.2835 Medline
42. S. V. Ramagopalan, A. Heger, A. J. Berlanga, N. J. Maugeri, M. R. Lincoln, A. Burrell, L.
Handunnetthi, A. E. Handel, G. Disanto, S. M. Orton, C. T. Watson, J. M. Morahan, G.
Giovannoni, C. P. Ponting, G. C. Ebers, J. C. Knight, A ChIP-seq defined genome-wide
map of vitamin D receptor binding: Associations with disease and evolution. Genome
Res. 20, 1352–1360 (2010). doi:10.1101/gr.107920.110 Medline
43. N. Ding, R. T. Yu, N. Subramaniam, M. H. Sherman, C. Wilson, R. Rao, M. Leblanc, S.
Coulter, M. He, C. Scott, S. L. Lau, A. R. Atkins, G. D. Barish, J. E. Gunton, C. Liddle,
M. Downes, R. M. Evans, A vitamin D receptor/SMAD genomic circuit gates hepatic
fibrotic response. Cell 153, 601–613 (2013). doi:10.1016/j.cell.2013.03.028 Medline
44. V. Nurminen, S. Seuter, C. Carlberg, Primary Vitamin D Target Genes of Human
Monocytes. Front. Physiol. 10, 194 (2019). doi:10.3389/fphys.2019.00194 Medline
45. M. Kawai, S. Kinoshita, M. Yamazaki, K. Yamamoto, C. J. Rosen, S. Shimba, K. Ozono, T.
Michigami, Intestinal clock system regulates skeletal homeostasis. JCI Insight 4,
e121798 (2019). doi:10.1172/jci.insight.121798 Medline
46. A. Hanel, A. Neme, M. Malinen, E. Hämäläinen, H.-R. Malmberg, S. Etheve, T.-P.
Tuomainen, J. K. Virtanen, I. Bendik, C. Carlberg, Common and personal target genes of
the micronutrient vitamin D in primary immune cells from human peripheral blood. Sci.
Rep. 10, 21051 (2020). doi:10.1038/s41598-020-78288-0 Medline
47. K. Litchfield, J. L. Reading, C. Puttick, K. Thakkar, C. Abbosh, R. Bentham, T. B. K.
Watkins, R. Rosenthal, D. Biswas, A. Rowan, E. Lim, M. Al Bakir, V. Turati, J. A.
Guerra-Assunção, L. Conde, A. J. S. Furness, S. K. Saini, S. R. Hadrup, J. Herrero, S.-H.
Lee, P. Van Loo, T. Enver, J. Larkin, M. D. Hellmann, S. Turajlic, S. A. Quezada, N.
McGranahan, C. Swanton, Meta-analysis of tumor- and T cell-intrinsic mechanisms of
sensitization to checkpoint inhibition. Cell 184, 596–614.e14 (2021).
doi:10.1016/j.cell.2021.01.002 Medline
48. M. L. Gjerstorff, The Danish Cancer Registry. Scand. J. Public Health 39, 42–45 (2011).
doi:10.1177/1403494810393562 Medline
49. J. F. H. Arendt, A. T. Hansen, S. A. Ladefoged, H. T. Sørensen, L. Pedersen, K. Adelborg,
Existing Data Sources in Clinical Epidemiology: Laboratory Information System
Databases in Denmark. Clin. Epidemiol. 12, 469–475 (2020).
doi:10.2147/CLEP.S245060 Medline
50. H. Quan, V. Sundararajan, P. Halfon, A. Fong, B. Burnand, J.-C. Luthi, L. D. Saunders, C.
A. Beck, T. E. Feasby, W. A. Ghali, Coding algorithms for defining comorbidities in
ICD-9-CM and ICD-10 administrative data. Med. Care 43, 1130–1139 (2005).
doi:10.1097/01.mlr.0000182534.19832.83 Medline
51. C.-B. Zhou, Y.-L. Zhou, J.-Y. Fang, Gut Microbiota in Cancer Immune Response and
Immunotherapy. Trends Cancer 7, 647–660 (2021). doi:10.1016/j.trecan.2021.01.010
Medline
52. C. Villemin, A. Six, B. A. Neville, T. D. Lawley, M. J. Robinson, G. Bakdash, The
heightened importance of the microbiome in cancer immunotherapy. Trends Immunol.
44, 44–59 (2023). doi:10.1016/j.it.2022.11.002 Medline
53. M. Alexander, P. J. Turnbaugh, Deconstructing Mechanisms of Diet-Microbiome-Immune
Interactions. Immunity 53, 264–276 (2020). doi:10.1016/j.immuni.2020.07.015 Medline
54. A. Sivan, L. Corrales, N. Hubert, J. B. Williams, K. Aquino-Michaels, Z. M. Earley, F. W.
Benyamin, Y. Man Lei, B. Jabri, M.-L. Alegre, E. B. Chang, T. F. Gajewski, Commensal
Bifidobacterium promotes antitumor immunity and facilitates anti-PD-L1 efficacy.
Science 350, 1084–1089 (2015). doi:10.1126/science.aac4255 Medline
55. M. Vétizou, J. M. Pitt, R. Daillère, P. Lepage, N. Waldschmitt, C. Flament, S. Rusakiewicz,
B. Routy, M. P. Roberti, C. P. M. Duong, V. Poirier-Colame, A. Roux, S. Becharef, S.
Formenti, E. Golden, S. Cording, G. Eberl, A. Schlitzer, F. Ginhoux, S. Mani, T.
Yamazaki, N. Jacquelot, D. P. Enot, M. Bérard, J. Nigou, P. Opolon, A. Eggermont, P.-L.
Woerther, E. Chachaty, N. Chaput, C. Robert, C. Mateus, G. Kroemer, D. Raoult, I. G.
Boneca, F. Carbonnel, M. Chamaillard, L. Zitvogel, Anticancer immunotherapy by
CTLA-4 blockade relies on the gut microbiota. Science 350, 1079–1084 (2015).
doi:10.1126/science.aad1329 Medline
56. B. Routy, E. Le Chatelier, L. Derosa, C. P. M. Duong, M. T. Alou, R. Daillère, A. Fluckiger,
M. Messaoudene, C. Rauber, M. P. Roberti, M. Fidelle, C. Flament, V. Poirier-Colame,
P. Opolon, C. Klein, K. Iribarren, L. Mondragón, N. Jacquelot, B. Qu, G. Ferrere, C.
Clémenson, L. Mezquita, J. R. Masip, C. Naltet, S. Brosseau, C. Kaderbhai, C. Richard,
H. Rizvi, F. Levenez, N. Galleron, B. Quinquis, N. Pons, B. Ryffel, V. Minard-Colin, P.
Gonin, J. C. Soria, E. Deutsch, Y. Loriot, F. Ghiringhelli, G. Zalcman, F. Goldwasser, B.
Escudier, M. D. Hellmann, A. Eggermont, D. Raoult, L. Albiges, G. Kroemer, L.
Zitvogel, Gut microbiome influences efficacy of PD-1–based immunotherapy against
epithelial tumors. Science 359, 91–97 (2018). doi:10.1126/science.aan3706 Medline
57. T. Tanoue, S. Morita, D. R. Plichta, A. N. Skelly, W. Suda, Y. Sugiura, S. Narushima, H.
Vlamakis, I. Motoo, K. Sugita, A. Shiota, K. Takeshita, K. Yasuma-Mitobe, D.
Riethmacher, T. Kaisho, J. M. Norman, D. Mucida, M. Suematsu, T. Yaguchi, V. Bucci,
T. Inoue, Y. Kawakami, B. Olle, B. Roberts, M. Hattori, R. J. Xavier, K. Atarashi, K.
Honda, A defined commensal consortium elicits CD8 T cells and anti-cancer immunity.
Nature 565, 600–605 (2019). doi:10.1038/s41586-019-0878-z Medline
58. V. Gopalakrishnan, C. N. Spencer, L. Nezi, A. Reuben, M. C. Andrews, T. V. Karpinets, P.
A. Prieto, D. Vicente, K. Hoffman, S. C. Wei, A. P. Cogdill, L. Zhao, C. W. Hudgens, D.
S. Hutchinson, T. Manzo, M. Petaccia de Macedo, T. Cotechini, T. Kumar, W. S. Chen,
S. M. Reddy, R. Szczepaniak Sloane, J. Galloway-Pena, H. Jiang, P. L. Chen, E. J.
Shpall, K. Rezvani, A. M. Alousi, R. F. Chemaly, S. Shelburne, L. M. Vence, P. C.
Okhuysen, V. B. Jensen, A. G. Swennes, F. McAllister, E. Marcelo Riquelme Sanchez,
Y. Zhang, E. Le Chatelier, L. Zitvogel, N. Pons, J. L. Austin-Breneman, L. E. Haydu, E.
M. Burton, J. M. Gardner, E. Sirmans, J. Hu, A. J. Lazar, T. Tsujikawa, A. Diab, H.
 Tawbi, I. C. Glitza, W. J. Hwu, S. P. Patel, S. E. Woodman, R. N. Amaria, M. A. Davies,
J. E. Gershenwald, P. Hwu, J. E. Lee, J. Zhang, L. M. Coussens, Z. A. Cooper, P. A.
Futreal, C. R. Daniel, N. J. Ajami, J. F. Petrosino, M. T. Tetzlaff, P. Sharma, J. P.
Allison, R. R. Jenq, J. A. Wargo, Gut microbiome modulates response to anti–PD-1
immunotherapy in melanoma patients. Science 359, 97–103 (2018).
doi:10.1126/science.aan4236 Medline
59. L. F. Mager, R. Burkhard, N. Pett, N. C. A. Cooke, K. Brown, H. Ramay, S. Paik, J. Stagg,
R. A. Groves, M. Gallo, I. A. Lewis, M. B. Geuking, K. D. McCoy, Microbiome-derived
inosine modulates response to checkpoint inhibitor immunotherapy. Science 369, 1481–
1489 (2020). doi:10.1126/science.abc3421 Medline
60. K. C. Lam, R. E. Araya, A. Huang, Q. Chen, M. Di Modica, R. R. Rodrigues, A. Lopès, S. B.
Johnson, B. Schwarz, E. Bohrnsen, A. P. Cogdill, C. M. Bosio, J. A. Wargo, M. P. Lee,
R. S. Goldszmid, Microbiota triggers STING-type I IFN-dependent monocyte
reprogramming of the tumor microenvironment. Cell 184, 5338–5356.e21 (2021).
doi:10.1016/j.cell.2021.09.019 Medline
61. C.-H. Lo, D.-C. Wu, S.-W. Jao, C.-C. Wu, C.-Y. Lin, C.-H. Chuang, Y.-B. Lin, C.-H. Chen,
Y.-T. Chen, J.-H. Chen, K.-H. Hsiao, Y.-J. Chen, Y.-T. Chen, J.-Y. Wang, L.-H. Li,
Enrichment of Prevotella intermedia in human colorectal cancer and its additive effects
with Fusobacterium nucleatum on the malignant transformation of colorectal adenomas.
J. Biomed. Sci. 29, 88 (2022). doi:10.1186/s12929-022-00869-0 Medline
62. N. Charoenngam, A. Shirvani, T. A. Kalajian, A. Song, M. F. Holick, The Effect of Various
Doses of Oral Vitamin D3 Supplementation on Gut Microbiota in Healthy Adults: A
Randomized, Double-blinded, Dose-response Study. Anticancer Res. 40, 551–556
(2020). doi:10.21873/anticanres.13984 Medline
63. P. Singh, A. Rawat, M. Alwakeel, E. Sharif, S. Al Khodor, The potential role of vitamin D
supplementation as a gut microbiota modifier in healthy individuals. Sci. Rep. 10, 21641
(2020). doi:10.1038/s41598-020-77806-4 Medline
64. C. E. Talsness, J. Penders, E. H. J. M. Jansen, J. Damoiseaux, C. Thijs, M. Mommers,
Influence of vitamin D on key bacterial taxa in infant microbiota in the KOALA Birth
Cohort Study. PLOS ONE 12, e0188011 (2017). doi:10.1371/journal.pone.0188011
Medline
65. S. Patrick, A tale of two habitats: Bacteroides fragilis, a lethal pathogen and resident in the
human gastrointestinal microbiome. Microbiology 168, 001156 (2022).
doi:10.1099/mic.0.001156 Medline
66. M. H. Sofi, Y. Wu, T. Ticer, S. Schutt, D. Bastian, H.-J. Choi, L. Tian, C. Mealer, C. Liu, C.
Westwater, K. E. Armeson, A. V. Alekseyenko, X.-Z. Yu, A single strain of Bacteroides
fragilis protects gut integrity and reduces GVHD. JCI Insight 6, e136841 (2021).
doi:10.1172/jci.insight.136841 Medline
67. Y. K. Lee, P. Mehrabian, S. Boyajian, W.-L. Wu, J. Selicha, S. Vonderfecht, S. K.
Mazmanian, The Protective Role of Bacteroides fragilis in a Murine Model of Colitis-
Associated Colorectal Cancer. MSphere 3, e00587-18 (2018).
doi:10.1128/mSphere.00587-18 Medline
68. J. A. Newton-Bishop, S. Beswick, J. Randerson-Moor, Y.-M. Chang, P. Affleck, F. Elliott,
M. Chan, S. Leake, B. Karpavicius, S. Haynes, K. Kukalizch, L. Whitaker, S. Jackson, E.
Gerry, C. Nolan, C. Bertram, J. Marsden, D. E. Elder, J. H. Barrett, D. T. Bishop, Serum
25-hydroxyvitamin D3 levels are associated with breslow thickness at presentation and
survival from melanoma. J. Clin. Oncol. 27, 5439–5444 (2009).
doi:10.1200/JCO.2009.22.1135 Medline
69. S. Muralidhar, A. Filia, J. Nsengimana, J. Poźniak, S. J. O’Shea, J. M. Diaz, M. Harland, J.
A. Randerson-Moor, J. Reichrath, J. P. Laye, L. van der Weyden, D. J. Adams, D. T.
Bishop, J. Newton-Bishop, Vitamin D–VDR Signaling Inhibits Wnt/β-Catenin–Mediated
Melanoma Progression and Promotes Antitumor Immunity. Cancer Res. 79, 5986–5998
(2019). doi:10.1158/0008-5472.CAN-18-3927 Medline
70. Ł. Galus, M. Michalak, M. Lorenz, R. Stoińska-Swiniarek, D. Tusień Małecka, A. Galus, T.
Kolenda, E. Leporowska, J. Mackiewicz, Vitamin D supplementation increases objective
response rate and prolongs progression-free time in patients with advanced melanoma
undergoing anti–PD-1 therapy. Cancer 129, 2047–2055 (2023). doi:10.1002/cncr.34718
Medline
71. T. Odamaki, J.-Z. Xiao, M. Sakamoto, S. Kondo, T. Yaeshima, K. Iwatsuki, H. Togashi, T.
Enomoto, Y. Benno, Distribution of different species of the Bacteroides fragilis group in
individuals with Japanese cedar pollinosis. Appl. Environ. Microbiol. 74, 6814–6817
(2008). doi:10.1128/AEM.01106-08 Medline
72. M. Barman, D. Unold, K. Shifley, E. Amir, K. Hung, N. Bos, N. Salzman, Enteric
salmonellosis disrupts the microbial ecology of the murine gastrointestinal tract. Infect.
Immun. 76, 907–915 (2008). doi:10.1128/IAI.01432-07 Medline
73. S. Zelenay, A. G. van der Veen, J. P. Böttcher, K. J. Snelgrove, N. Rogers, S. E. Acton, P.
Chakravarty, M. R. Girotti, R. Marais, S. A. Quezada, E. Sahai, C. Reis e Sousa,
Cyclooxygenase-Dependent Tumor Growth through Evasion of Immunity. Cell 162,
1257–1270 (2015). doi:10.1016/j.cell.2015.08.015 Medline
74. A. Faustino-Rocha, P. A. Oliveira, J. Pinho-Oliveira, C. Teixeira-Guedes, R. Soares-Maia, R.
G. da Costa, B. Colaço, M. J. Pires, J. Colaço, R. Ferreira, M. Ginja, Estimation of rat
mammary tumor volume using caliper and ultrasonography measurements. Lab Anim. 42,
217–224 (2013). doi:10.1038/laban.254 Medline
75. A. Cardoso, A. Gil Castro, A. C. Martins, G. M. Carriche, V. Murigneux, I. Castro, A.
Cumano, P. Vieira, M. Saraiva, The Dynamics of Interleukin-10-Afforded Protection
during Dextran Sulfate Sodium-Induced Colitis. Front. Immunol. 9, 400 (2018).
doi:10.3389/fimmu.2018.00400 Medline
76. J. A. McCulloch, J. H. Badger, N. Cannon, R. R. Rodrigues, M. Valencia, J. J. Barb, M. R.
Fernandes, A. Balaji, L. Crowson, C. O’hUigin, A. Dzutsev, G. Trinchieri, JAMS - A
framework for the taxonomic and functional exploration of microbiological genomic
data. bioRxiv 2023.03.03.531026 [Preprint] (2023);
https://doi.org/10.1101/2023.03.03.531026.
77. C. N. Spencer, J. L. McQuade, V. Gopalakrishnan, J. A. McCulloch, M. Vetizou, A. P.
Cogdill, M. A. W. Khan, X. Zhang, M. G. White, C. B. Peterson, M. C. Wong, G. Morad,
T. Rodgers, J. H. Badger, B. A. Helmink, M. C. Andrews, R. R. Rodrigues, A. Morgun,
 Y. S. Kim, J. Roszik, K. L. Hoffman, J. Zheng, Y. Zhou, Y. B. Medik, L. M. Kahn, S.
Johnson, C. W. Hudgens, K. Wani, P.-O. Gaudreau, A. L. Harris, M. A. Jamal, E. N.
Baruch, E. Perez-Guijarro, C.-P. Day, G. Merlino, B. Pazdrak, B. S. Lochmann, R. A.
Szczepaniak-Sloane, R. Arora, J. Anderson, C. M. Zobniw, E. Posada, E. Sirmans, J.
Simon, L. E. Haydu, E. M. Burton, L. Wang, M. Dang, K. Clise-Dwyer, S. Schneider, T.
Chapman, N. A. S. Anang, S. Duncan, J. Toker, J. C. Malke, I. C. Glitza, R. N. Amaria,
H. A. Tawbi, A. Diab, M. K. Wong, S. P. Patel, S. E. Woodman, M. A. Davies, M. I.
Ross, J. E. Gershenwald, J. E. Lee, P. Hwu, V. Jensen, Y. Samuels, R. Straussman, N. J.
Ajami, K. C. Nelson, L. Nezi, J. F. Petrosino, P. A. Futreal, A. J. Lazar, J. Hu, R. R. Jenq,
M. T. Tetzlaff, Y. Yan, W. S. Garrett, C. Huttenhower, P. Sharma, S. S. Watowich, J. P.
Allison, L. Cohen, G. Trinchieri, C. R. Daniel, J. A. Wargo, Dietary fiber and probiotics
influence the gut microbiome and melanoma immunotherapy response. Science 374,
1632–1640 (2021). doi:10.1126/science.aaz7015 Medline
78. J. A. McCulloch, D. Davar, R. R. Rodrigues, J. H. Badger, J. R. Fang, A. M. Cole, A. K.
Balaji, M. Vetizou, S. M. Prescott, M. R. Fernandes, R. G. F. Costa, W. Yuan, R.
Salcedo, E. Bahadiroglu, S. Roy, R. N. DeBlasio, R. M. Morrison, J.-M. Chauvin, Q.
Ding, B. Zidi, A. Lowin, S. Chakka, W. Gao, O. Pagliano, S. J. Ernst, A. Rose, N. K.
Newman, A. Morgun, H. M. Zarour, G. Trinchieri, A. K. Dzutsev, Intestinal microbiota
signatures of clinical response and immune-related adverse events in melanoma patients
treated with anti-PD-1. Nat. Med. 28, 545–556 (2022). doi:10.1038/s41591-022-01698-2
Medline
79. B. Li, C. N. Dewey, RSEM: Accurate transcript quantification from RNA-Seq data with or
without a reference genome. BMC Bioinformatics 12, 323 (2011). doi:10.1186/1471-
2105-12-323 Medline
80. A. Subramanian, P. Tamayo, V. K. Mootha, S. Mukherjee, B. L. Ebert, M. A. Gillette, A.
Paulovich, S. L. Pomeroy, T. R. Golub, E. S. Lander, J. P. Mesirov, Gene set enrichment
analysis: A knowledge-based approach for interpreting genome-wide expression profiles.
Proc. Natl. Acad. Sci. U.S.A. 102, 15545–15550 (2005). doi:10.1073/pnas.0506580102
Medline