Abstract
Islet transplantation offers a promising treatment for type 1 diabetes
(T1D), by aiming to restore insulin production and improve glycemic
control. However, T1D is compounded by impaired angiogenesis and immune
dysregulation, which hinder the therapeutic potential of cell
replacement strategies. To address this, this work evaluates the
proangiogenic and immunomodulatory properties of mesenchymal stem cells
(MSCs) to enhance vascularization and modulate early‐stage immune
rejection pathways in the context of islet allotransplantation. This
work employs the Neovascularized Implantable Cell Homing and
Encapsulation (NICHE) platform, a subcutaneous vascularized implant
with localized immunomodulation developed by the group. This study
assesses vascularization and immune regulation provided by MSCs, aiming
to improve islet survival and integration in diabetic rats while
considering sex as a biological variable. These findings demonstrate
that MSCs significantly enhance vascularization and modulate the local
microenvironment during the peri‐transplant period. Importantly, this
work discovers sex‐specific differences in both processes, which
influence islet engraftment and long‐term function.
Keywords: immunomodulation, islet engraftment, mesenchymal stem cells,
sex‐specific differences, Type 1 diabetes, vascularized subcutaneous
microenvironment
__________________________________________________________________
With emphasis on a diabetic state, this study highlights the role of
mesenchymal stem cells (MSCs) in enhancing vascularization and
modulating immune responses within a subcutaneous platform for cell
transplantation for type 1 diabetes (T1D). Comprehensive profiling of
MSC‐induced angiogenic and immune responses, provides insight into the
role of MSCs in a vascularized subcutaneous system for islet
transplantation.
graphic file with name ADVS-12-2411574-g007.jpg
1. Introduction
Type 1 diabetes (T1D) is a chronic metabolic disorder characterized by
insulin deficiency resulting from autoimmune destruction of
insulin‐producing pancreatic cells. Endothelial and immune
dysregulation in individuals with T1D leads to vascular aberrations and
chronic inflammation,^[ [62]^1 ^] increasing the susceptibility of
developing long‐term comorbidities. These comorbidities include
diabetic angiopathy, nephropathy, neuropathy, and cardiovascular
diseases,^[ [63]^2 ^] among other complications that impinge on the
efficacy of T1D management. Transplantation of insulin‐producing cells
offers a viable strategy to manage T1D via cell replacement. However,
pre‐existing vascular and immune abnormalities in a diabetic setting^[
[64]^3 ^] can reduce the efficacy of cell therapies, as transplanted
cells require a highly oxygenated microenvironment with immediate
access to nutrients. One effective approach is to provide a
well‐vascularized engraftment site, which necessitates protection from
immune rejection.
To this end, owing to their pivotal role in vascular regeneration and
immunomodulation,^[ [65]^4 ^] mesenchymal stem cells (MSCs) have been
leveraged as accessory cells for T1D cell replacement therapy.^[ [66]^5
^] The low expression of class II MHC molecules renders them
hypoimmunogenic,^[ [67]^6 ^] which is conducive for co‐delivery with
allogeneic cell therapies. Further, MSC can inhibit T cell
proliferation, promote anti‐inflammatory macrophage polarization^[
[68]^7 ^] and induce transplant tolerance.^[ [69]^8 ^] Preclinically,
co‐culture with MSC has resulted in improved pancreatic islet viability
and insulin secretion.^[ [70]^9 ^] Moreover, MSCs have been shown to
induce long‐term graft acceptance when administered alone or in
combination with short‐term immunosuppression (IS) treatments.^[
[71]^10 ^] Of relevance, ongoing clinical studies are assessing the
efficacy of islets and MSC co‐transplantation for diabetic and chronic
pancreatitis patients,^[ [72]^11 ^] some of which have demonstrated
β‐islet restoration and amelioration of hyperglycemia.^[ [73]^12 ^]
Distinct from previous studies, here we leveraged a diabetic setting to
evaluate the proangiogenic and immunomodulatory capacity of MSCs in the
context of islet allotransplantation. We focused our analysis on the
subcutaneous space as its accessibility offers an attractive
alternative to the clinical standard of portal vein infusion.^[ [74]^13
^] In this setting, the challenges associated with limited
vascularization of the subcutaneous space are exacerbated by angiopathy
associated with the diabetic state. Thus, we investigated whether MSCs
could enhance vascularization and overcome the vascular impairment in
the diabetic condition. Further, we aimed to delineate the capacity of
MSCs in immunomodulating early‐stage immune rejection pathways
activated upon allogeneic islet transplantation. Importantly, in the
context of both vascularization^[ [75]^14 ^] and immunomodulation,^[
[76]^15 ^] preclinical and clinical studies have evidenced
pathophysiological differences in male versus female individuals. As
such, central to our study, we considered sex as a key biological
variable.
In this study we used a subcutaneous cell therapy platform, the
Neovascularized Implantable Cell Homing and Encapsulation (NICHE),
previously developed in our group.^[ [77]^16 ^] Conceived for the
transplantation of therapeutic cells,^[ [78]^16a,c ^] the NICHE
provides a defined vascularized 3D tissue compartment that is engrafted
within the subcutaneous space,^[ [79]^16b,d,e ^] allowing for free
immune cell trafficking and cytokine and chemokine transport. In
essence, the NICHE enables reproducible assessment of MSC local
function via minimally invasive graft monitoring or defined tissue
retrieval and provides a localized setting for studying spatiotemporal
immune responses to allogeneic transplants and the therapeutic
modulation thereof.
Specifically, our investigation includes the systematic and
longitudinal assessment of both vascularization and immunomodulation
against allogeneic islet transplants provided by MSCs in an
immunocompetent rat model of diabetes. We profiled the NICHE local
immune microenvironment after allogeneic pancreatic islet
transplantation and evaluated the immunomodulating properties of MSCs
in mitigating early immune responses post‐transplant. Additionally, we
investigated the efficacy of MSCs as a supportive therapy for
supporting islet engraftment.
2. Results
2.1. NICHE Integration and Vascularization Impairment in Diabetic Rats
The integration of NICHE implanted within subcutaneous tissues was
previously studied in healthy animals. In that setting, NICHE was
determined to be biocompatible, and enhanced engraftment and
subcutaneous vascularization were observed with the use of MSCs loaded
within the NICHE cell reservoir.^[ [80]^16a–c ^] In this study, we
sought to assess whether the foreign body response (FBR) to and
vascularization of NICHE (Figure [81]1A) is affected by the
proinflammatory state in diabetic rats, where increased recruitment of
neutrophils and proinflammatory macrophages^[ [82]^17 ^] can cause
device integration impairment (Figure [83]1B; Figure [84]S1A,
Supporting Information). For this, immunocompetent rats were rendered
diabetic using streptozotocin (STZ) prior to subcutaneous implantation
of NICHE, with glycemic control maintained via insulin‐releasing
pellets (Linplant) (Figure [85]1C,D). Implant reactivity and
vascularization were compared with non‐diabetic (healthy) rats^[
[86]^16a ^] used as controls. After 6 weeks of subcutaneous
implantation, the fibrotic capsules around the NICHE in diabetic
animals had lax, vascularized, and significantly thicker fibrotic
capsules (222.5 ± 72.30 µm) than immunocompetent healthy rats (128.9 ±
38.99 µm) (p < 0.05, Figure [87]1E–G). Comparatively, within diabetic
animals, males had thicker fibrotic capsules than females (p < 0.01,
Figure [88]S1B, Supporting Information). Reactivity scoring used to
gauge cellular responses to the NICHE showed that diabetic rats
exhibited higher tissue reactivity, evidenced by increased immune cell
infiltration and tissue inflammation, compared to healthy rats
(Figure [89]1H). However, there were no sex‐specific differences in
reactivity scores (Figure [90]S1C, Supporting Information). Further,
the NICHE cell reservoirs in diabetic animals had complete tissue
penetration across the entire cell reservoir (Figure [91]1J),
comparable to that of healthy rats (Figure [92]1I). However, the
diabetic cohort had significantly lower blood vessel area (3.3%;
Figure [93]1K) and number (37.8 ± 20.2 vessels mm^−2; Figure [94]1L)
than healthy rats (6.1% and 341 ± 148.2 vessels mm^−2). Collectively,
our results indicate a stronger pro‐inflammatory response to the
implant and impaired angiogenesis in the NICHE subcutaneous
microenvironment in diabetic animals as compared to healthy controls.
Figure 1.
Figure 1
[95]Open in a new tab
NICHE implant reactivity and vascularization differences between
healthy and diabetic rats. A) Schematic of implanted NICHE device and
B) longitudinal cross‐section of subcutaneous local microenvironment.
C) Experimental design of vascularization study in Fisher (F344) rats.
STZ = streptozotocin. D) BG measurements of male and female diabetic
rats implanted with NICHE and Linplant (n = 25 to day 14, n = 16 to day
28, n = 8 to day 42); and healthy male rats (n = 3) for reference.
Horizontal dotted line indicates glycemic control threshold.
Representative Masson's trichrome staining of fibrotic capsule around
NICHE implanted in E) healthy and F) diabetic male rats for 6 weeks.
Scale bars, 200 µm. G) Quantification of fibrotic capsule thickness and
H) implant reactivity scores of NICHE implanted in healthy (n = 4) and
diabetic (n = 4) rats for 6 weeks. Mean ± SD, un‐paired Student's
t‐test (*p < 0.05). Vascularized cell reservoir tissue sections stained
with B. simplicifolia lectin (BS‐1) in I) healthy and J) diabetic rats.
Scale bars, 50 µm. K) Area of tissue comprised by blood vessels and L)
vessels quantification of NICHE implanted in healthy (n = 3) and
diabetic (n = 4) rats (n = 8–10 technical replicates each). Mean ± SD
of averaged technical replicates, un‐paired Student's t‐test (*p <
0.05, ***p < 0.001).
2.2. MSCs Support NICHE Subcutaneous Vascularization in Diabetic Rats
Next, we explored MSC‐driven vascularization in a time‐ and
sex‐dependent manner in diabetic rats. We subcutaneously implanted
MSC‐hydrogel loaded NICHE (MSC) in STZ‐induced diabetic male and female
rats and assessed vascularization over a period of 6 weeks, compared to
control hydrogel‐only implants. Animals were maintained under glycemic
control via exogenous insulin therapy to recapitulate a clinical
therapeutic scenario and preserve their well‐being.
We used lectin staining to define the circumference of blood vessels
(Figure [96]S1D, Supporting Information) and calculate total blood
vessel area, as well as determine the number of vessels. At 2 weeks
post‐implantation, the male MSC group showed complete development of
tissue in the cell reservoir compared to the control, with larger blood
vessel area (5.6% versus 4.3%, n.s.) and higher number of vessels
(38.52 ± 16.05 vessels mm^−2 versus 23.66 ± 5.6 vessels mm^−2, p = 0.1)
(Figure [97]2A–F). In females, the MSC group had significantly larger
blood vessel area (4.1% versus 2.2%, p < 0.05) and similar vessel
density (32.19 ± 12.47 vessels mm^−2 versus 33.59 ± 6.35 vessels mm^−2,
n.s.) compared to control at 2‐weeks post‐implantation
(Figure [98]2O–T).
Figure 2.
Figure 2
[99]Open in a new tab
NICHE subcutaneous vascularization in diabetic rats. Representative
histological images of H&E‐stained cross sections of explanted NICHE
and 20× magnification of NICHE cell reservoir sections with blood
vessels stained in red with BS‐1. H&E and BS‐1‐stained sections of
control devices 2‐week post‐implantation for A, B) males and O, P)
females. Violin plots of C, Q) area of tissue comprised by blood
vessels and D, R) number of vessels inside cell reservoirs throughout 6
weeks after implantation quantified from BS‐1‐stained sections in males
and females, respectively (n = 4–5 samples per condition; n = 6–10
fields of view). Violin plots show all captured fields of view, two‐way
ANOVA of averaged FOV values (*p < 0.05, **p < 0.01). 2‐week
post‐implantation sections of NICHE loaded with MSCs in E, F) males and
S, T) females. Control devices with H&E and BS‐1 for G, H) males and U,
V) females; MSC devices with H&E and BS‐1 sections for K, L) males and
Y, Z) females 4 weeks post‐implantation. Control devices with H&E and
BS‐1 for I, J) males and W, X) females; MSC devices with H&E and BS‐1
sections for M, N) males and AA, BB) females implanted for 6 weeks.
Scale bars in H&E, 1 mm (left) and in BS‐1, 100 µm (right).
By week 4 post‐implantation, MSC groups (Figure [100]2K,L,Y,Z) showed a
significant increase in vascularization with a larger total area
covered by blood vessels (6.8%, p < 0.05) compared to controls in male
rats (Figure [101]2G,H). Likewise, a larger number of vessels (60.8 ±
15.27 vessels mm^−2, p < 0.05) was observed in females
(Figure [102]2R), when compared to female controls (Figure [103]2U,V).
By week 6, the blood vessel area and number of vessels in control
groups remained unchanged with respect to week 4 in both male and
female rats (Figure [104]2I,J,W,X). At this timepoint, MSC groups
showed a significant decrease in the area covered by blood vessels
(3.5%, p < 0.01) in males (Figure [105]2M,N) and reduced blood vessel
density (31.93 ± 9.8 vessels mm^−2, p < 0.01) in females
(Figure [106]2AA,BB) compared to week 4.
Collectively, we noted that male animals showed a larger area of tissue
occupied by blood vessels, whereas females had higher number of blood
vessels but not total blood vessel area, indicative of smaller blood
vessels in females. Overall, we demonstrate that MSCs supported
subdermal vascularization in male and female diabetic animals, with the
strongest angiogenic response observed at 4‐weeks post‐implantation,
followed by a decrease thereafter.
2.3. MSC Induction of Functional Vasculature in the NICHE Subcutaneous
Microenvironment
Vascular integrity within the subcutaneous microenvironment is key for
subdermal graft revascularization. MSC‐mediated angiogenesis is driven
by vascular endothelial growth factor (VEGF) secretion.^[ [107]^18 ^]
Therefore, we quantified VEGF protein levels in the cell reservoir
tissue of control and MSC‐loaded NICHE devices explanted at 2‐, 4‐, and
6‐weeks post‐implantation. Our results indicate evident VEGF secretion
at 2 weeks, followed by a marked decline at later timepoints in
MSC‐loaded devices in both male and female rats (Figure [108]3A,B).
Then, we further assessed the effect of MSCs in achieving a mature
functional network in the NICHE microenvironment of diabetic male
(Figure [109]3C) and female (Figure [110]3D) rats. We used CD31, eNOS,
and VE‐Cadherin as markers of mature blood vessels: CD31 is a cell
adhesion and signaling protein that is expressed on the surface of
endothelial cells^[ [111]^19 ^] and serves as an indicator of vascular
structures. Endothelial nitric oxide synthase (eNOS) and vascular
endothelial cadherin (VE‐Cadherin) are markers of vessel maturity,
function, and integrity.^[ [112]^20 ^]
Figure 3.
Figure 3
[113]Open in a new tab
MSC induction of functional vasculature development in NICHE.
Quantification of VEGF in the cell reservoir of control (n =
3–4/timepoint) and MSC‐loaded (n = 5/timepoint) NICHE devices implanted
for 2, 4, and 6 weeks in A) males and B) females. Protein levels were
normalized to total protein content of the tissue homogenates. Mean ±
SD, two‐way ANOVA with Bonferroni's multiple comparisons test (*p <
0.05; **p < 0.01; ***p < 0.001). Representative immunofluorescent
staining of NICHE vasculature in C) males and D) females at 4 weeks
post‐implantation stained with functional blood vessel markers CD31
(red), eNOS (gold) and VE‐Cadherin (magenta). RBCs are autofluorescent
in FITC channel (green). Scale bars, 50 µm. Fluorescence intensity
analysis of E,H) CD31, F,I) eNOS, and G,J) VE‐Cad as relative
expression in NICHE‐MSC compared to control devices at each timepoint
for males and females (n = 4 biological replicates; n = 4 fields of
view per sample). Scatter plots show mean of all captured FOV (n = 16),
un‐paired Student's t‐test of averaged FOV per sample at each timepoint
(*p < 0.05, **p < 0.01, ***p < 0.001), denoting level of significance
compared to control hydrogel only‐NICHE.
At 2 weeks post‐implantation, MSC female group showed eightfold higher
expression of CD31 than control hydrogel only (Figure [114]3D,H), in
accordance with the larger blood vessel area (Figure [115]2Q).
Thereafter, CD31 expression decreased but was still significantly
higher (3.5‐fold) than control at 4 weeks post‐implantation. In males,
at weeks 4 and 6, MSC group had a sixfold and sevenfold increase in
CD31 expression, respectively, with respect to control
(Figure [116]3E). MSC groups showed higher expression of eNOS, with a
significant increase of threefold and fourfold relative to control, at
weeks 2 and 4, respectively, in males (Figure [117]3F). In females,
eNOS was significantly increased in the MSC group (fivefold) compared
to control at week 4 (Figure [118]3I). Increase in VE‐Cad expression
was modest, compared to eNOS and CD31 in the MSC‐NICHE groups; however,
there was a significant twofold increase in males at week 6 and a
fivefold increase in females at week 4 (Figure [119]3G,J).
In addition, the presence of patent blood vessels connected to systemic
circulation was confirmed with the presence of autofluorescent red
blood cells (RBCs) in the lumen (Figure [120]3C,D). Moreover, blood
vessels were permeable to fluorescently labeled 10 kDa dextran with
consistent extravasation in both groups, while larger 70 kDa dextran
was better retained in the lumen of penetrating blood vessels in the
MSC groups (Figure [121]S2, Supporting Information). Taken together,
this data indicated that MSCs can promote mature and functional
vasculature in a diabetic setting without any signs of differentiation
in the subcutaneous microenvironment (Figures [122]S3–S5, Supporting
Information). Maximal enhancement of vessel density and maturity was
achieved between 4‐ and 6‐weeks post‐implantation for both sexes.
Therefore, in the following studies, a 5‐week pre‐vascularization
period was considered optimal to maintain balance of a
well‐vascularized space without regression of functional blood vessels.
2.4. MSCs Promote Syngeneic Islet Engraftment in the NICHE Vascularized
Microenvironment
We further explored the effect of MSCs in supporting islet engraftment
and revascularization in a diabetic and syngeneic model. Diabetic rats
received a subtherapeutic dose of syngeneic islets with or without MSC
co‐transplantation in the NICHE cell reservoir at 5 weeks
post‐implantation. To minimize competition for oxygen and nutrients, we
employed a 2:1 islet to MSC ratio based on our previous studies.^[
[123]^16d ^] After explant at 1‐ and 4‐weeks post‐transplantation, the
tissue within NICHE devices underwent a process of clarification via
delipidation^[ [124]^21 ^] and was imaged via lightsheet microscopy.
Lectin and insulin immunofluorescence staining allowed for 3D
visualization of vascular architecture and pancreatic islet
distribution, respectively, within NICHE (Figure [125]4A–H). By week 1,
we observed patent vasculature within the tissue and blood vessels in
the vicinity of pancreatic islets with (Figure [126]4A,B) or without
MSCs (Figure [127]4E,F). This vasculature remained constant on week 4,
with increased intra‐islet lectin signal in both groups
(Figure [128]4C,D,G,H). Islet engraftment, calculated as the percentage
of total islet volume nominally loaded in each device, was higher in
MSC‐co‐transplanted islets compared to islets alone on days 7
(48.45% versus 38.9%, p = 0.4) and 28 (63.66% versus 27.35%, p = 0.01)
(Figure [129]4I). However, no differences in the fractional blood
vessel volume (Figure [130]4J) and intra‐islet capillary volume
(Figure [131]4K) were observed across groups. Taken together, these
data support the properties of MSCs in enhancing islet engraftment and
potentially improving transplant outcomes in the NICHE device.
Figure 4.
Figure 4
[132]Open in a new tab
Effect of MSCs on islet engraftment and revascularization. EZ Clear
processed, wholemount lightsheet fluorescent microscopy imaged NICHE
cell reservoir of islet‐loaded devices explanted at days A,B) 7 and
C,D) 28; and islets + MSC‐loaded devices explanted at days E,F) 7 and
G,H) 28. Islets are stained with insulin‐Alexa Flour 555 (green) and
blood vessels are labeled with fluorescently conjugated Lycopersicon
esculentum lectin (lectin‐DyLight 649). Scale bars; 1 mm (Top), 400 µm
(bottom). I) Islet survival calculated as % of total islet volume
loaded in NICHE. Blood vessel volume analysis of J) total cell
reservoir and K) intra‐islet volume of no MSC (n = 3/timepoint) and MSC
co‐transplanted (n = 3 /timepoint) devices. Mean ± SD, two‐way ANOVA
with Bonferroni's multiple comparison (*p < 0.05).
2.5. Immunomodulatory Effect of MSCs on Islet Allograft Survival in Diabetic
Rats
We next assessed the immunomodulatory effect of MSCs in the context of
protecting allogeneic pancreatic islets from acute immune rejection
(Figure [133]5A). First, we examined the in vitro cytocompatibility of
Fisher‐derived MSCs with Lewis rat donor islets, using islets alone as
a control. Live/dead imaging analysis showed no difference in islet
viability (Figure [134]S6A,B, Supporting Information), and insulin
secretion response was not affected in islets cocultured with MSCs
(Figure [135]S6C, Supporting Information). Additionally, there was no
difference in apoptotic (Annexin^+/PI^−) and necrotic (Annexin^−/PI^+)
cells after 3 days in culture with MSCs, when compared to control
(Figure [136]S6D–G, Supporting Information).
Figure 5.
Figure 5
[137]Open in a new tab
Immunomodulatory effect of MSCs for allogeneic islet transplantation in
immunocompetent and diabetic rats. A) Study design. STZ =
streptozotocin, allo‐tx = allogeneic transplant. BG measurements of B)
male and C) female F344 diabetic rats receiving collagen injection
(control; n = 12 to day 3, n = 8 to day 7, n = 4 to day 14),
islets‐only (islets), or islets co‐transplanted with MSCs (islets +
MSC) in NICHE cell reservoir. (Islets, Islets + MSC; n = 14 to day 3, n
= 10 to day 7, n = 5 to day 14). #: comparison of control versus
islets, &: comparison of control versus islets + MSC. Horizontal dotted
line indicates glycemic control threshold. Weight tracking of D) male
and E) female rats. Mean ± SEM, one‐way ANOVA with Tukey's multiple
comparisons test (# or & p < 0.05; ## or && p < 0.01; ### or &&& p <
0.001). Blood glucose AUC for F) day 0 to day 3 (control, n = 12;
islets and islets + MSC, n = 14), for G) day 0 to day 7 (control, n =
8; islets and islets + MSC, n = 10), and for H) day 0 to day 14
(control, n = 4; islets and islets + MSC, n = 5). Mean ± SD, Two‐way
ANOVA followed by Tukey's multiple comparisons test (* p < 0.05, ** p <
0.01, *** p < 0.001). IMC of cell reservoir from I) control, J) islets,
and K) islets + MSC devices. Scale bars, 100 µm. Cell population
quantification for L) males and M) females. (n = 3–5/group), mean ± SD,
one‐way ANOVA with Tukey's multiple comparisons test per population (*p
< 0.05; **p < 0.01; ***p < 0.001).
Next, MSC‐loaded NICHE were subcutaneously implanted in STZ‐induced
diabetic male and female Fisher rats for 5‐weeks of
pre‐vascularization. Thereafter, rats were randomized into three
groups: 1) “Islets + MSC” in which allogeneic Lewis rat islets were
co‐transplanted with syngeneic MSCs into the NICHE cell reservoir; 2)
“Islets,” which was allogeneic islets only; and 3) “Vehicle,” which was
the same collagen hydrogel used in the other groups to transplant
islets and MSCs. The latter group served both as control and baseline
response to the transcutaneous injection. Glycemic control was
maintained in all groups with insulin‐releasing pellets from day 35 up
to days 14 and 7 prior to islet transplant, for males and females,
respectively (Figure [138]S7A,B, Supporting Information). The starting
blood glucose (BG) at the time of transplant was 490.9 ± 24.81 mg dL^−1
for male and 385.2 ± 30.31 mg dL^−1 for female rats (Figure [139]5B,C).
By day 7 after transplantation, the BG in males significantly decreased
to 366.3 ± 63.38 mg dL^−1 and 459.9 ± 57.53 mg dL^−1 in Islets + MSC
group and Islets cohort, respectively, compared to control (573.4 ±
42.48 mg dL^−1; p < 0.05). By day 14, the BG of male rats receiving
islets‐only returned to pre‐transplant levels (555.6 ± 67.81 mg dL^−1),
whereas those co‐transplanted with MSCs maintained similar values to
that of day 7 (371.6 ± 81.88 mg dL^−1).
In contrast, on day 7, the BG of females dropped in islets + MSC and
islets‐only groups to 215.2 ± 48.47 and 213.4 ± 26.46 mg dL^−1,
respectively, which was close to the levels of healthy animals
(200 mg dL^−1). The female control rats had significantly higher BG
levels of 540.5 ± 73.47 mg dL^−1 (p < 0.001, both groups). On day 14,
BG levels of female islet cohorts remained significantly lower than
control (642 ± 36.95; Figure [140]5C). Specifically, the female islets
+ MSC group had BG levels of 225.8 ± 46.6 mg dL^−1, which was not
significantly different than that observed on day 7 (p = 0.89).
Instead, the BG on day 14 of islets‐only group increased by 30% to
281.6 ± 88.21 mg dL^−1, suggesting islet rejection. Moreover, the area
under the curve (AUC) of blood glucose levels of male rats
co‐transplanted with MSCs was significantly reduced compared to
islets‐only group at days 3, 7, and 14 (Figure [141]5F–H). In contrast,
the AUC in females was significantly reduced in both experimental
groups compared to control, but not different between both groups.
These results suggest that MSCs in the NICHE can mitigate the acute
immune rejection, and this effect was more notable in males.
In addition, weight was tracked throughout the study as an indicator of
well‐being following diabetes induction. The animals showed healthy
weight gain during pre‐vascularization due to glycemic control (Figure
[142]S7C,D, Supporting Information). However, weight stalled around the
time of transplant for control animals, while rats receiving islets
exhibited weight increase (Figure [143]5D,E). Particularly, the male
islets + MSC group had a significant increase in weight only at day 7
(p = 0.008) and the female islets + MSC rats showed a significant
increase at study endpoint (p = 0.04).
Further, we evaluated the local transplant microenvironment in a
spatial manner at day 7 post‐transplant as a midway point of acute
immune rejection. We characterized immune cells spatial distribution
and interaction via imaging mass cytometry (IMC) analysis. Unsupervised
clustering of IMC data classified cell populations into 15 distinct
clusters (Figure [144]S8A, Supporting Information). Control devices
showed minimal presence of cytotoxic T‐cells (CD8^+) and macrophages
(CD68^+), along with vascular structures with α‐smooth muscle actin
(αSMA^+) (Figure [145]5I). In the islet‐containing groups, pancreatic
islets were identified by the presence of beta (insulin^+) and alpha
(glucagon^+) cells, and their distribution was consistent with islet
morphology (Figure [146]5J,K). Transplanted male animals exhibited
significant infiltration of CD8^+ T‐cells, independent of MSC
co‐transplant, and higher than that observed in females. However,
significant macrophage infiltration was only noted in the male
islets‐only group (Figure [147]5J,L). Notably, only female rats
co‐transplanted with MSCs showed a significant presence of alpha cells
(glucagon^+) and a higher density of regulatory T‐cells (Treg, 306
cells mm^−2) compared to islets‐only group (15.5 cells mm^−2, p = 0.09)
(Figure [148]5K,M). Additionally, the subcellular spatial resolution
obtained from IMC allowed for analysis of tissue architecture and
cell‐cell interactions. Single‐cell phenomapping allowed us to perform
a correlation analysis between the density of insulin^+ and glucagon^+
cells and other cell populations (Figure [149]S8, Supporting
Information). The proportion of CD4^+ and CD8^+ T‐cells, CD68^+MHCII^+,
and GranzymeB^+ cells in close proximity to islet cells was similar
between the islets‐only and islets + MSC groups. However, we observed a
shift from a negative to a positive correlation between islet cells and
both Treg (Figure [150]S8H, Supporting Information) and αSMA^+ clusters
(Figure [151]S8I, Supporting Information) in the islets + MSC group.
These results suggest that MSC co‐transplantation is associated with
the preservation of Treg within the transplant microenvironment and
promotes allogeneic islet revascularization. This is a relevant
finding, considering that the animals were immunocompetent, and no
induction immunosuppression was used.
2.6. Immune Characterization of the Transplant Microenvironment in NICHE
To investigate the immunomodulatory effect of MSCs in mitigating early
immunological events post‐transplantation, the NICHE local
microenvironment was profiled via mass cytometry by time‐of‐flight
(CyTOF). We analyzed infiltrating immune cell populations in NICHE with
islets + MSC and islets‐only, compared to control. T‐distributed
stochastic neighbor embedding (tSNE) plots showed that cell
infiltration is initiated by macrophages (CD68^+) and dendritic cells
(DCs, CD68^−) on days 3 and 7 for both sexes, which is also evident in
the control (Figure [152]6A–C,E). Macrophages were comprised of two
CD68^+ subpopulations, CD11b^+CD11c^− or CD11c^+CD11b^−. There was a
slight predominance of CD11b^+ macrophages at day 3 post‐transplant and
a shift to CD11c^+ macrophages after 7 days, which was more apparent in
the females. More specifically, in both sexes, co‐transplantation with
MSCs induced higher CD11b^+ macrophages on day 3. The presence of M1
and M2 macrophages was increased with MSC co‐transplantation in males
on days 7 and 14 (Figure [153]6D). In contrast, MSCs reduced the levels
of M1 and M2 macrophages in females on days 3 and 7. On day 14, the
female islets + MSC cohort showed an increase in M2 macrophages
(Figure [154]6F). Further, in both sexes, DC infiltration on day 3
remained unchanged compared to control; however, this effect was
reversed in females on day 7. Overall, the initial presence of myeloid
cells across groups was similar (Figure [155]S9E–H, Supporting
Information). We attribute this to the transcutaneous injection method
used to load the cell reservoir, which contributes to initial
inflammation.
Figure 6.
Figure 6
[156]Open in a new tab
Immune characterization in NICHE local microenvironment after
allogeneic islet transplant. tSNE plots of immune cells infiltrating
the NICHE cell reservoir for A) males and B) females throughout 14 days
post‐transplant. Heatmap of log[2] fold change in immune cell
populations as percent of CD45^+ cells in C) males and E) females;
heatmap of log[2] fold change in immune cell subpopulations as percent
of parent in D) males and F) females, with respect to control group.
Cell population frequencies (%) shown in Figure [157]S9, Supporting
Information. Quantification of VEGF and TGFβ in the cell reservoir of
NICHE devices implanted in G,H) males and I,J) females (n = 3–5
group/timepoint). Protein levels were normalized to total protein
content of the tissue homogenates. Mean ± SD, two‐way ANOVA with
Tukey's multiple comparisons test (*p < 0.05; **p < 0.01; ***p <
0.001). Quantification of immunomodulatory cytokines in the
peri‐transplant period showing expression relative to control group in
K) males and L) females. Mean ± SD, two‐way ANOVA (*p < 0.05; **p <
0.01). Concentration values (pg mL^−1) shown in Figure [158]S10,
Supporting Information.
We noted a gradual infiltration of CD4^+ and CD8^+ T cells over time in
response to islet transplantation, particularly on day 14 for both
sexes. On day 3, when MSCs were co‐transplanted with islets, both sexes
showed decreased CD4^+ and CD8^+ T cells. This effect was particularly
notable in females, where the islets‐only cohort had a 33% increase in
CD8^+ T cell infiltration, whereas the addition of MSCs showed a
reduction relative to control (Figure [159]6E). Additionally, the
addition of MSCs in females showed an increase in Tregs, limited to day
3. By day 7, in males, MSC co‐transplantation was associated with an
increase in CD4^+ T cells and a reduction in CD8^+ T cell populations
(Figure [160]6C). Further evaluation of CD4^+ subpopulations in this
male cohort showed an increase in effector memory cells (Tem) and
decrease in central memory cells (Tcm). Within the CD8^+ population in
males, Tem and Tcm as well as granzyme B^+ cells were reduced with MSC
co‐transplantation (Figure [161]6D). In females on day 7, MSC
co‐transplantation decreased CD4^+ T cells compared to the islets‐only
group, contrasting the response seen in males. Specifically, the islets
+ MSC female group showed an increase in cytotoxic T‐lymphocyte
associated antigen‐4 (CTLA4) expression on Tregs. On day 14, the
increase in CD4^+ T cells persisted in males with MSC
co‐transplantation, with no notable differences in their
subpopulations, except for a reduction in Tregs. Additionally, by day
14, there were no notable differences in the CD4^+ and CD8^+ T cells
and their subpopulations in females, except for an increase in CTLA4^+
Tregs in the islets + MSC group (Figure [162]6F, Figure [163]S9T,
Supporting Information). As expected, T‐cell migration occurs gradually
over time, eventually leading to allograft destruction within weeks,
with MSCs providing only mild protection or delaying this process.
Furthermore, MSC co‐transplantation decreased B cell and neutrophil
infiltration, evident on day 3 in both sexes. Male rats co‐transplanted
with MSCs showed 27% fewer infiltrating B cells and 83% less
neutrophils than islets‐only cohort (Figure [164]6C, Figure [165]S9A,B,
Supporting Information). Similarly, female rats co‐transplanted with
MSCs had 60% and 75% fewer infiltrating B cells and neutrophils,
respectively, compared to islets‐only (Figure [166]6E, Figure
[167]S9C,D, Supporting Information). Additionally, at day 14,
MSC‐transplanted groups had 40% less infiltration of natural killer
(NK) cells compared to islets only, evident in both sexes (Figure
[168]S9J,L, Supporting Information). NK cells are key initiators of the
rejection cascade, and MSCs significantly reduced their proportion.
Collectively, this data indicates that MSCs were able to modulate the
early immune response during allogeneic islet transplantation in
immunocompetent animals. However, their effect was transient, as they
were unable to overcome the allogeneic response driven by T cells and
the adaptive immunity by day 14.
2.7. Local Immune Secretome Analysis
We used VEGF and transforming growth factor beta (TGF‐β) levels as
surrogates for MSC‐mediated angiogenic and immunomodulatory signaling,
respectively, in the NICHE microenvironment.^[ [169]^22 ^]
Co‐transplantation with MSCs significantly increased VEGF concentration
(62.2 ± 18.1 pg mg^−1, p < 0.001) in males (Figure [170]6G) and (18.47
± 5.5 pg mg^−1, p < 0.01) in female rats (Figure [171]6I) with respect
to control at day 3 post‐transplant. Additionally, increased VEGF was
also significant when compared to islets‐only group (27.8 ± 12.7
pg mg^−1, p < 0.05) in males only. VEGF concentration decreased over
time, suggesting a transient expression or migration of MSCs out of the
local microenvironment. At day 3, TGF‐β levels were significantly
increased in MSCs co‐transplanted males (Figure [172]6H) and females
(Figure [173]6I) compared to islets‐only and control groups, suggesting
direct secretion by MSCs. To further investigate the immunomodulatory
function of MSCs in situ, we quantified anti‐ and pro‐inflammatory
cytokines in the NICHE microenvironment. Pro‐tolerogenic cytokines IL‐4
and IL‐2 showed a slight increase in rats co‐transplanted with MSCs in
the early acute period. More specifically, males had a 26% increase in
IL‐4 on day 7, whereas females had a 39% increase on day 3. IL‐2 levels
at day 3 post‐transplant showed a significant 125% increase in female
rats co‐transplanted with MSCs (Figure [174]6K,L). Pro‐inflammatory
monocyte chemoattractant protein‐1 (MCP‐1), responsible for immune cell
recruitment during immunological rejection of cellular transplants, was
decreased in rats co‐transplanted with MSCs. Specifically,
co‐transplantation with MSCs reduced the levels of MCP‐1 by 40% in
males on day 3. In line with this, IL‐6 levels were increased in
cohorts receiving islets‐only, but not in MSC co‐transplanted groups
(Figure [175]S10, Supporting Information). Additionally, IFN‐γ
accumulation in the NICHE microenvironment significantly increased over
time and at day 14 post‐transplant, transplanted rats with islets‐only
showed significant increase with respect to control in males
(Figure [176]6K), and with respect to the MSC‐co‐transplanted group in
females (Figure [177]6L). These results indicate that MSC
co‐transplantation enhances early angiogenic and immunosuppressive
signaling, reducing inflammatory responses in the NICHE local
microenvironment.
2.8. Effect of MSCs on Draining Lymph Node and Systemic Immune Response
Following a tissue transplant, the draining lymph node (dLN) becomes a
focal point for the initiation and regulation of the adaptive immune
response.^[ [178]^23 ^] To this end, we assessed the dLN and spleen
during the acute rejection period. In the dLN, we observed an increase
in T cells (Figure [179]7A,E), particularly evident on days 3 and 7 in
the male islets‐only cohort. For CD8^+ T cells, we observed a
statistically significant increase on day 7 in both islets‐only
cohorts, which was subdued by the addition of MSCs (Figure [180]7B,F).
Moreover, CD8^+ T cells in the dLN of female islets‐only group remained
increased until day 14, whereas MSCs sustained levels comparable to
that of the control. For the islet‐only male and female cohorts, there
were no significant changes in Treg cells at the dLN across the
different time points (Figure [181]7C,G). However, female islets + MSC
group displayed increased Treg on day 7 when compared to control and
islets‐only cohorts (Figure [182]7G, p = 0.1), and the ratio of Treg to
CD8^+ T cells was significantly higher than the islets‐only
(Figure [183]7D,H). The immunosuppressive effect of MSCs on T cell
proliferation was further assessed via carboxyfluorescein succinimidyl
ester (CFSE) dilution assay using lymphocytes harvested from the dLN at
day 7 post‐transplant. MSCs displayed proliferative inhibition of T
cells within CFSE‐labeled lymphocytes cocultured with irradiated
splenocytes from donor rats (Figure [184]S10, Supporting Information).
Moreover, Tregs were preserved when MSCs were present in the mixed
lymphocyte reaction (Figure [185]S10B,F, Supporting Information). This
could be of relevance as Tregs are shown to suppress naïve T cell
proliferation in the dLN in allograft studies.^[ [186]^24 ^]
Figure 7.
Figure 7
[187]Open in a new tab
Characterization of immune response at dLN. Flow cytometry data
represented as % of A,E) CD3^+ and B,F) CD8^+ T cells in CD45^+ cells,
and C,G) Treg cells in CD4^+ T cells in males and females. D,H) Ratio
of Treg to CD8^+ T cells in males and females, respectively. (All
groups, n = 4–5 per timepoint). Histogram plots of infiltrating
CD11b^−CD11c^+ DCs across 14 days post‐transplant and quantitative
analysis of %DCs in CD45^+ cells for I,J) males and K,L) females.
Histogram plots of infiltrating CD11b^+CD68^+ macrophages and
quantitative analysis of %macrophages in CD45^+ cells for M,N) males
and O,P) females. (All groups, n = 4–5 per timepoint). Mean ± SD,
two‐way ANOVA with Fisher's LSD test (*p < 0.05, **p < 0.01, ***p <
0.001).
The islets‐only cohorts displayed a statistically significant increase
in DCs (CD11b^−CD11c^+) compared to control on day 7 in the dLN of both
male (Figure [188]7I,J and p < 0.01) and female (Figure [189]7K,L and p
< 0.05) rats. Augmented macrophage migration to the dLN was observed
until day 14 in male islets‐only rats (Figure [190]7M,N), while this
response was not apparent in females (Figure [191]7O,P). Overall, the
effects of the allogeneic transplant on T cells and antigen‐presenting
cell (APC) populations at the dLN were mostly noticeable 7 days after
the loading procedure.
In the spleen, no changes in T cells, APCs and macrophages populations
were observed across experimental groups, indicating that the
transplant did not impact the systemic immune homeostasis over the
14‐day period (Figure [192]S11, Supporting Information). Taken
together, these data indicate that after transplanting allogeneic
pancreatic islets into the NICHE cell reservoir, the innate rejection
cascade is initiated and further amplified in the dLN but remains
confined within the local and focal microenvironment without triggering
a systemic immune response. Additionally, the anti‐inflammatory effect
of MSCs was limited at the dLN site without inducing systemic effects.
2.9. MSC‐Modulation Promotes a Dynamic Tissue Architecture in the NICHE
Microenvironment
While MSCs have been implicated in a wide variety of immunomodulation
strategies, including organ and cell transplantation,^[ [193]^25 ^]
their modulatory mechanism, particularly when deployed in the
subcutaneous microenvironment, remains poorly elucidated. To address
this, we used Visium HD spatial transcriptomics (Figure [194]8A) to
investigate the spatial localization of gene signatures for
transplanted islets and their surrounding microenvironment when
co‐transplanted with MSCs. Unsupervised clustering analysis of Visium
HD datasets identified 8–12 distinct cell clusters per sample
(Figure [195]8B) within the region of interest (ROI), defined by
examination of H&E staining (Figure [196]8C). To further harmonize
clustering across samples for deeper genomic analysis, we performed
t‐SNE analysis, defining 3 meta‐clusters based on shared gene
expression profiles (Figure [197]8B). H&E‐defined pancreatic islet
morphology was consistent with expression of common pancreatic islet
markers (Figure [198]8D), which allowed for identification of clusters
consisting mainly of islet cells and their surrounding microenvironment
(Figure [199]8E). Identified clusters were merged into meta‐clusters
representing major tissue types with shared gene expression
(Figure [200]8F). Additionally, the presence of MSC‐related modulatory
markers was detected in the defined ROI (Figure [201]8G), suggesting
local persistence of the MSCs that were co‐transplanted with islets.
Figure 8.
Figure 8
[202]Open in a new tab
Integrative overview of the MSC‐modulated NICHE transplant
microenvironment. A) Schematic workflow of Visium HD spatial
transcriptomics of FFPE sections of NICHE local microenvironment after
islet + MSC co‐transplant (n = 2 males, n = 2 females). B) tSNE
analysis followed by principal‐component analysis (PCA) of gene
expression matrixes from identified clusters (n = 41) in all samples (n
= 4). C) Representative H&E‐stained image of the NICHE local
microenvironment. Scale bar, 1 mm. D) Normalized gene expression
(normalized by total UMI counts) of pancreatic islet markers in
selected region of interest (ROI). Scale bar, 200 µm. E) Spatial plot
of Visium HD clusters identified with Loupe Browser before and F) after
meta‐cluster assignment. G) Spatial gene expression plot of modulatory
markers related to MSCs in selected ROI. Scale bar, 200 µm. Volcano
plots of differentially expressed genes highlighting upregulated genes
(log[2]FC > 0.05 and ‐log[10](p‐value) ≥ 1.5) in red for H) islet, I)
islet periphery, and J) vascularized tissue meta‐clusters. Gene
Ontology (GO) pathways associated with upregulated signature genes for
K) islet, L) islet periphery, and M) vascularized tissue meta‐clusters.
Regulated pathways identified with normalized enrichment score (NES) >1
and FDR < 0.1. RT‐PCR analysis of islet, vascular, regulatory and
MSC‐related genes in NICHE cell reservoir tissues with allogeneic
islets only or islets + MSC explanted at 7 days post‐transplant from N)
male and O) female diabetic rats. Gene expression was normalized to
Gapdh. Fold changes in gene expression are relative to islets only
groups. (n = 3–6 biological replicates per group), mean ± SD, unpaired
Student's t‐test (*p < 0.05) for each assayed gene. n.d. = not
detected.
Up‐regulated genes (Log[2]FC >1, FDR < 0.05) were identified to define
molecular signature profiles of islets, peri‐islet and vascularized
niche meta‐clusters (Figure [203]8H–J). The top up‐regulated genes in
the islet signature showed structural, functional, and developmental
markers (i.e., Iapp, Cltrn, Vegfa, Mafa, Ins2, Gcg, Nkx2‐2, and
Neurod1). The peri‐islet upregulated genes included transcriptional,
immune, and signaling markers such as Ccna2, Bcl6, Cdk1, Il12b, CD44,
Irf1, Tgfb1, Col5a3, and Vim. The remaining vascularized tissue
signature genes are related to tissue homeostasis, collagen formation,
and cell migration, such as Igf1, Gpc3, Lrp6, Tnxb, Fbn2, Col1a1, and
Cxcl12, also known as stromal cell‐derived factor 1 (SDF‐1). Notably,
MSCs spontaneously produce SDF‐1, which promotes the mobilization of
endothelial progenitor cells, key contributors to VEGF production.^[
[204]^26 ^]
To further investigate the specific functional differences between
molecular signatures, gene set enrichment analysis (GSEA) was
performed, and relevant differential pathways were elucidated. The top
20 upregulated pathways in the islet signature were mainly involved in
insulin secretion, endocrine function, and blood vessel regulation
(Figure [205]8K). Upregulation of the different channel activities
reflects the high metabolic state of healthy, functional islets. The
peri‐islet upregulated genes were immune related, particularly immune
activation, cytokine response, and T cell receptor signaling.
Immunoregulatory pathways were also significantly enriched, such as
regulation of CD4^+ T cell which involves both T‐helper and regulatory
cell differentiation. In addition, VEGFA‐VEGFR2 and MAPK signaling were
upregulated, suggesting angiogenic activity in the islet periphery
(Figure [206]8L). The vascularized NICHE tissue showed upregulation of
Wnt signaling, extracellular matrix organization, angiogenesis, and
transcriptional programs associated with growth factor signaling and
leukocyte migration (Figure [207]8M), indicative of tissue organization
of the vascularized cell reservoir.
Finally, we validated differences in gene expression of a subset of
islet, vascular, regulatory, inflammatory, and MSC‐related genes at day
7 post‐transplant in MSC co‐transplanted grafts compared to grafts
receiving islet only (Figure [208]8N,O). Co‐delivery with MSCs induced
a significant decrease of pro‐inflammatory Ccl2 in males
(Figure [209]8N) and Ifng in females (Figure [210]8O). Moreover, MSC
co‐transplantation induced an increase in vascular Angpt2, albeit this
was quantifiable in females only. In females, MSC co‐transplantation
induced an increase in Nt5e which was undetectable in males. NT5E can
act as an inhibitory immune checkpoint molecule, as the production of
free adenosine suppresses cellular immune responses.^[ [211]^27 ^]
Collectively, these results demonstrate the dynamic processes modulated
by MSCs in the NICHE microenvironment.
3. Discussion
The limited vascularization in the subcutaneous space presents a
challenge for islet transplantation, especially considering that T1D is
associated with endothelial dysfunction and impaired angiogenesis. Here
we leveraged the subcutaneous NICHE islet encapsulation platform^[
[212]^16a ^] to assess the angiogenic and immunomodulatory functions of
MSCs in the context of allogeneic islet transplantation in diabetic
rats, without additional immunosuppressive therapy. We comprehensively
profiled the MSC‐modulated subcutaneous transplant microenvironment
using integrative, multiplexed approaches including 3D imaging, CyTOF,
chemokine quantification, and spatial analyses of protein and gene
expression; our results revealed some sex‐specific differences in both
the microenvironment and transplant outcomes. Our study aims to provide
more insight into the role of MSCs in an allogenic subcutaneous
transplant setting and a rationale for complementary IS regimens that
could promote a tolerogenic state.
Previously, we demonstrated that the FBR in concert with MSCs promoted
vascularization into the NICHE cell reservoir in healthy rats.^[
[213]^16a,b,e ^] In this study, we observed that diabetic rats have
impaired vascularization, thicker fibrotic encapsulation, and higher
tissue reactivity to the NICHE when compared to healthy counterparts
(Figure [214]1H). This is consistent with defective diabetic wound
healing and angiogenesis, which involves prolonged inflammation,
oxidative stress, reduced growth factor production, and altered immune
cell responses.^[ [215]^17 , [216]^28 ^] Here we showed that MSCs
promoted angiogenesis and enhanced mature and functional vasculature
within the NICHE ≈4–6 weeks post implantation in diabetic hosts,
similar to the time frame in healthy animals.^[ [217]^16a,b ^] VEGF
production by MSCs was detected at 2‐weeks post‐implantation,
suggesting their proangiogenic activity is induced during the early
hypoxic conditions before the development of functional blood vessels
within the cell reservoir (Figure [218]3). Further, despite reduced
vascularization compared to healthy animals, diabetic rats with
MSC‐loaded NICHE showed improved blood vessel area, number, and
functionality compared to control. Notably, MSCs stimulated a larger
number of blood vessels, albeit smaller in size, in diabetic female
rats, compared to males. This is consistent with known sex‐specific
differences, where females tend to have smaller blood vessels than
males.^[ [219]^29 ^]
We showed neovessel integration with the systemic circulation and islet
revascularization by 7 days post‐transplantation (Figure [220]4), both
of which are crucial for successful islet engraftment. Increased VEGF
levels during the 14‐day period in rats co‐transplanted with MSC
alludes to their angiogenic activity (Figure [221]6G,I). However, we do
not preclude the fact that increased VEGF production may also result
from MSC‐induced mobilization of endothelial progenitor cells,^[
[222]^26 ^] monocyte recruitment,^[ [223]^30 ^] or from stress
responses from hypoxic islets.^[ [224]^31 ^] Furthermore, while
vascularization is critical for engraftment, it also enables immune
cell trafficking, facilitating rejection of transplanted cells.
Consistent with this, immune cell characterization of the NICHE
microenvironment revealed an early influx of B cells, neutrophils, DCs
and macrophages by day 3 following allogeneic islet loading. By day 14,
there was markedly increased T cell infiltration. Therefore, successful
engraftment requires effective modulation of both innate and adaptive
immune responses to abrogate graft rejection.^[ [225]^32 ^] To date,
there is no standard IS regimen for clinical islet transplantation
(CIT), and localized immunomodulation is an emerging strategy that
remains to be fully realized.^[ [226]^32 , [227]^33 ^] Ideally, an
immunomodulated transplant microenvironment has high Treg levels for
tolerance, along with low local counts of cytotoxic T cells^[ [228]^34
^] and M1 pro‐inflammatory macrophages.^[ [229]^35 ^]
Given their potential role as immunomodulatory accessory cells to
improve islet survival,^[ [230]^5b–f ^] we evaluated MSC
co‐transplantation as a local adjuvant strategy for allogeneic islet
transplantation. Our results showed that MSCs exert a transient
immunosuppressive effect, where increased TGF‐β levels at day 3 likely
promoted an anti‐inflammatory environment and facilitated a regulatory
T cell phenotype,^[ [231]^36 ^] This is supported by reduced
infiltration of B cells, macrophages, and neutrophils
(Figure [232]6C,E) and increased Treg density in the NICHE,
particularly in female rats on days 3 and 7 post‐transplant
(Figure [233]5M, Figure [234]6F, Figure [235]S8H, Supporting
Information). Moreover, an increase in CTLA4^+ Tregs in females
receiving islets co‐transplanted with MSCs suggests that these cells
may dampen T cell activation by outcompeting CD28 for binding to
CD80/CD86 on APCs,^[ [236]^37 ^] MSC co‐transplantation increased IL‐2
and IL‐4, and decreased IFN‐γ levels in the NICHE, consistent with
established regulatory mechanisms.^[ [237]^38 ^] These findings are
consistent with prior data showing that MSCs can prolong islet
allograft survival via Treg recruitment.^[ [238]^5c ^] In vitro studies
have shown that MSCs can inhibit NK cell proliferation and their
associated IFN‐γ production.^[ [239]^39 ^] Here, the MSC‐associated
reduction of intra‐graft IFN‐γ concentration at days 7 and 14
(Figure [240]6K,L) was consistent with reduced Ifng gene expression at
day 7 post‐transplant (Figure [241]8O), and diminished infiltration of
NK cells at day 14. This suggests an early suppressive effect by MSCs
that becomes apparent at later timepoints. However, while MSCs
modulated the early innate immune response, the T cell driven adaptive
response on day 14 was not suppressed. Nevertheless, MSC
co‐transplantation reduced T cell and APC migration to the dLN
(Figure [242]7N,P), confirming the localized immunomodulatory effect.
Overall, these findings suggest that MSCs contribute to improved islet
survival during the peri‐transplant period by modulating the early
immune responses.
Spatially resolved gene expression analysis of the subcutaneous
vascularized transplant microenvironment in the NICHE offers insights
into the tissue architecture of this prevascularized device strategy
for islet transplantation. Our findings suggest that MSCs can regulate
fibroblast activity and limit excessive collagen production, thereby
facilitating cell migration, tissue remodeling, and continuous immune
cell influx. Moreover, the peri‐islet associated gene signature
predominantly reflects transcriptional programs involved in both immune
activation and regulation, suggesting that infiltrated immune cells
reside in close proximity to the islets. This spatial gene expression
assessment reveals that upregulated pathways involved in pancreatic
islet function, immune responses, angiogenesis, cell migration, and
tissue remodeling are predominant in our MSC‐modulated system.
Our study shows that MSCs provide an initial immunomodulatory benefit
for islet survival. However, the inherent migratory nature of MSCs^[
[243]^40 ^] limits their long‐term local persistence. Repeated MSC
dosing could boost their activity,^[ [244]^41 ^] although the cost and
feasibility of multiple administrations in a clinical scenario would
pose a challenge. In this context, alternative strategies such as the
use of MSC spheroids to enhance cell retention and survival and boost
their local immunomodulatory effects^[ [245]^42 ^] are being explored.
Combining MSC co‐transplantation with targeted immunosuppressive
regimens may offer a synergistic approach to prolong graft viability.
Such approach would entail optimizing combinations of immunosuppressive
agents administered either before or concurrently with cell
transplantation. The selection of immunosuppressive agents to be
strategically combined with MSCs must not interfere with the reported
pro‐inflammatory signals necessary for MSC activation.^[ [246]^43 ^]
For instance, rapamycin can modulate T cell responses while sparing
regulatory T cell function,^[ [247]^44 ^] thereby potentially
synergizing with MSC‐mediated immune regulation. Moreover,
costimulatory blockade agents such as CTLA‐4Ig and anti‐CD40L, have
shown promise in reducing alloreactive T cell activation^[ [248]^45 ^]
and may promote a tolerogenic environment after MSC priming. To this
end, the NICHE platform offers the opportunity to explore targeted
immunomodulation via local co‐delivery of islets, immunomodulatory
cells, and immunosuppressive agents in a confined setting.
The limited availability of studies exploring sex‐specific differences
in islet transplantation underscores the need to consider these
variables in future research. Our findings revealed notable differences
in vascularization and transplant outcomes between male and female
rats. In females, a lower degree of vascularization may slow immune
cell infiltration and trafficking to lymphoid tissues, thereby delaying
allograft rejection. Furthermore, lower baseline blood glucose levels
at the time of transplant are likely to favorably influence the
glycemic response following islet transplantation, as it is known to
significantly impact islet engraftment and overall transplant
outcomes.^[ [249]^46 ^] Hormonal influences may also play a role, as
estrogen improves insulin sensitivity.^[ [250]^47 ^] These sex‐specific
differences are difficult to portray at the gene expression level. A
possible explanation is that the dynamic physiological factors driving
vascularization and therapeutic outcomes, such as blood flow, immune
cell trafficking, and modulatory mechanisms, are not fully captured by
static gene expression snapshots.^[ [251]^48 ^] As we move toward
individualized therapeutic approaches, integrating sex‐specific
variations into treatment strategies may inform optimal MSC dosing for
enhancing both vascularization and immunomodulation, while emphasizing
the critical role of managing BG before islet transplantation.
Limitations of our study include the following aspects: The STZ‐induced
diabetic rat model used in this study does not replicate the
autoimmunity observed in T1D, which is involved in the immune
destruction of cell allografts and affects vascularization. Due to
NICHE size constraints, we could not use NOD mice, a more
representative model for T1D, and existing autoimmune rat models are
limited by low diabetes incidence,^[ [252]^49 ^] making them unsuitable
for our purposes. Therefore, we opted for the STZ‐diabetic rat model,
aware of the inability to account for autoimmunity effects. Further,
long‐term engraftment was not explored due to the absence of induction
IS, which, while necessary to prevent acute rejection, it would have
hindered our understanding of MSC‐driven effects. Additionally, we used
commercially available bone‐marrow derived MSCs, which require an
invasive harvesting procedure in a clinical setting, posing potential
complications, particularly for diabetic patients. Therefore, further
investigation is needed to determine whether allogeneic MSCs can match
the clinical efficacy of autologous sources. Moreover, not all results
reached statistical significance, making it difficult to draw broad
definitive conclusions. However, several notable overall trends support
the reported angiogenic and immunomodulatory effects of MSCs.
Additionally, while the use of histological sections limited
vascularization quantification across the entire specimen, this was
addressed through the use of lightsheet microscopy, which enabled the
comprehensive visualization of the blood vessel network. Last, the
permanence of MSCs in the vascularized NICHE microenvironment was not
directly assessed, which would entail long‐term in vivo cell tracking
via IVIS, MRI, or intravital microscopy imaging. However, our results
suggest MSC retention is limited to the first week post‐transplant.
4. Conclusion
In summary, our study demonstrates the proangiogenic and
immunomodulatory properties of MSCs in the context of the subcutaneous
transplantation of islet allograft in diabetic rats. Our analysis,
focused on a diabetic setting in both males and females, provides
valuable insight for the future development of MSC‐based adjuvant
strategies for the delivery of therapeutic cells. Furthermore, our
findings support the rationale for combining site‐specific
immunomodulatory approaches in vascularized subcutaneous systems to
improve islet transplantation outcomes.
5. Experimental Section
NICHE Fabrication
NICHE design was scaled down compared to our previous publications
(18.9 mm × 15.4 mm × 3.8 mm versus 30.4 mm × 15.4 mm × 3.8 mm)^[
[253]^16a,c,d,f ^] to allow for safe implantation in diabetic female
rats, which are significantly smaller than their male counterpart.
NICHE main components remained the same, with a U‐shaped drug reservoir
surrounding the central cell reservoir, enclosed in two‐layered nylon
meshes. MSCs and islets were transplanted in the cell reservoir,
whereas the drug reservoirs were unloaded. Fabrication and assembly
protocols remained similar to our previous work. Briefly, the structure
of the device was designed in Solidworks (Dassault Systèmes) and
3D‐printed with biocompatible polyamide (PA2200, EOS). It was fitted
with two nanoporous polyether‐sulfone membranes (30 nm, Sterlitech) and
two pairs of nylon meshes (outer: 100 µm, inner: 300 µm; Elko
Filtering), all affixed using implantable‐grade silicone adhesive
(MED3‐4213, Nusil). The loading and refilling ports were created using
the same silicon adhesive. All components underwent autoclaving before
assembly in a sterile setting within a laminar flow hood. The assembled
NICHE devices were gas sterilized using ethylene oxide at the Current
Good Manufacturing Practice (cGMP) core at Houston Methodist Research
Institute (HMRI) facility.
Animal Models
8‐week‐old male and female Fischer (CDF) (F344; Charles River Strain
Code 002, MHC Haplotype RT1^lv) rats were used for diabetes induction
and then throughout the study. Animals were housed in the Comparative
Medicine Program facility at Houston Methodist Research Institute under
controlled environmental conditions (12 h light/dark cycle) and
maintained in pairs with free access to water and Teklad Global 18%
protein diet (Envigo). Daily assessments included blood glucose (BG)
measurements and welfare checks. Humane endpoints were defined by signs
including marked lethargy, hypothermia, severe dehydration, hematuria,
ataxia, hunched posture, labored breathing, abnormal gait, implant site
infection, wound dehiscence, body weight loss exceeding 20%, body
condition score (BCS) below 2, and severe hypoglycemia. At study
endpoints, animals were euthanized via isoflurane overdose. All animal
protocols were approved by the Houston Methodist Institutional Animal
Care and Use Committee (IACUC, #IS00007362) and were conducted
following the NIH Guide for the Care and Use of Laboratory Animals, PHS
Animal Welfare Policy, and the Animal Welfare Act (original protocol
detailed in previous publication^[ [254]^16a ^]).
Biocompatibility Assessment
Biocompatibility was assessed in STZ‐induced diabetic rats implanted
subcutaneously with NICHE for 6 weeks and compared to age‐matched
healthy rats. Following explantation, the implants were fixed in 10%
formalin for 3 days followed by ethanol dehydration and processed for
histology. Fibrotic capsule thickness was quantified on Masson's
Trichrome‐stained tissue sections with QuPath (v0.5.1). Technical
replicates (n = 8) were averaged, and biological replicates (n = 4)
were pooled for data visualization. Implant reactivity was evaluated in
H&E‐stained sections. A board‐certified pathologist, blinded to
treatment groups, performed the histological scoring using a previously
published system.^[ [255]^50 ^]
Implantation of NICHE in Diabetic Rats
8‐week‐old male and female Fisher rats were rendered diabetic via
intraperitoneal (IP) streptozotocin (STZ; CAS 18883‐66‐4, Sigma)
injection of 50 mg kg^−1 single dose. BG measurements were taken daily,
and diabetes was confirmed with 3 consecutive readings >300 mg dL^−1.
Subcutaneous fluids were provided daily using the fluid replacement
Equation [256]1, until day of implantation.
[MATH: Fluidvolumeml=Bodyweightg×%dehydrationasdecimal<
/mtr> :MATH]
(1)
Six days after STZ diabetes induction, rats were implanted
subcutaneously with NICHE device and a 3 mm long insulin releasing
pellet (Linplant, Linshin Canada) for glycemic control. For
implantation, rats received buprenorphine injection of 1 uL g^−1 of
body weight, 2 h prior to surgery. For vascularization experiments,
sterile NICHE devices were loaded with Control vehicle hydrogel (20%
Pluronic F‐127 in DMEM) or bone marrow MSCs (RAFMX‐01001, Cyagen, Lot.
210330H61) resuspended in vehicle hydrogel (5 ( 10^5 cells per NICHE).
Implantation surgery procedure was performed as previously described.^[
[257]^16a ^] Briefly, after sedation with 2% isoflurane, a 2 cm
incision was made to create two subcutaneous pockets, into which NICHE
devices were inserted on each side of the rat dorsum. The incision was
then closed with wound‐clips.
BG Monitoring and Supportive Care
BG levels were monitored by tail prick using a commercial veterinary
glucometer (AlphaTrack III, Zoetis) with the assigned canine code for
the test strips. BG was monitored daily after diabetes induction and
until 10 days after insulin pellet implantation, then it was monitored
every other day. For vascularization experiments, rats were
re‐implanted with Linplant to continue therapy when hyperglycemia
recured. For immunomodulation experiments, no insulin pellets were
re‐implanted to evaluate islet transplant therapeutic effect.
Hyperglycemia (3 consecutive BG readings >300 mg dL^−1) recured at
least 7 days prior to islet transplant. BG was monitored daily after
islet transplant. The area under the BG curve was computed for
individual animals and averaged within groups. The calculation included
all the timepoints between day 0 and day 3 (control, n = 12; islets and
islets + MSC, n = 14), day 0 and day 7 (control, n = 8; islets and
islets + MSC, n = 10), or between day 0 and day 14 (control, n = 4;
islets and islets + MSC, n = 5). Calculated total areas were used for
comparison between groups.
Hypoglycemic episodes were controlled with subcutaneous administration
of 1 mL Lactated Ringer's and 5% Dextrose fluids (Baxter) when BG was
50–60 mg dL^−1 with depressed mentation. Additionally, when BG levels
were <50 mg dL^−1, IP fluids were provided along with heat support
until improved mentation or BG readings were observed. All animals
received nutritional support to maintain weight, including Nutra Gel
Complete Nutrition (Bio‐Serv), Supreme Mini‐Treats (Bio‐Serv), and
moistened pellets of their regular diet.
Vascularization Assessment
At weeks 2, 4, and 6 post‐implantation, NICHE devices along with
surrounding tissue were surgically removed at study termination. The
explanted tissues were fixed in 10% formalin for 3 days followed by
ethanol dehydration and processed for histology. Prior to paraffin
embedding, the NICHE framework was removed from the fixed tissue to
allow sectioning.
Tissue sections (5 µm) were stained with hematoxylin‐eosin (H&E) and
Masson's Trichrome (MT) at the HMRI Research Pathology Core. For blood
vessel labeling, antigen retrieval was performed by boiling slides with
rodent decloaker solution (RD913M, Biocare Medical) in pressure cooker
for 20 min cycle and cooled on bench‐top for 30 min. Slides were then
blocked with 5% normal goat serum in 0.1% bovine serum albumin/tris
buffered saline (BSA/TBS) for 1hr at room temperature. This was
followed by overnight incubation at 4 °C with biotinylated B.
simplicifolia lectin (L3759, Sigma, 10 µg mL^−1) and subsequent 30 min
incubation at RT with streptavidin AP (434 322, Invitrogen, 1:100).
Sections were developed with Warp Red Chromogen system (5 083 328,
Biocare Medical) and counterstained with hematoxylin and Tacha's bluing
solution following manufacturer's instructions. Whole‐slide scans and
magnified fields of view (FOV) were obtained with Keyence BZ‐X800
Microscope (Keyence) with 10× and 20× objectives, respectively. Eight
to ten FOV were randomly captured from each slide, and blood vessels
were quantified by a blinded evaluator. Vessel density was determined
as the number of blood vessels per mm^2 using Equation ([258]2) and
vessel area was calculated using Equation ([259]3) as previously
described.^[ [260]^16a,b ^]
[MATH: Bloodvesseldensity=VesselnumberFOVarea
:MATH]
(2)
[MATH: Vesselarea%=AreaoccupiedbyvesselsTotalsectionareax100 :MATH]
(3)
For immunofluorescence staining, following deparaffinization,
rehydration, pressure‐cooker‐based antigen retrieval, and blocking with
5% goat serum in 0.1% BSA/TBS, sections were incubated with primary
antibodies CD31 (NB100‐2284, Novus Biologicals, 1:200), VE Cadherin
(36‐1900, Invitrogen, 1:25), and eNOS (ab300071, Abcam, 1:50) diluted
in 1% BSA, 1% horse serum, 0.3% TritonX‐100, and 0.01% sodium azide in
1× PBS. Secondary anti‐rabbit Alexa Flour 555 antibody (A‐21428,
Invitrogen, 1:200) was then applied. Prolong mountant with NucBlue was
added to preserve fluorescence ([261]P36981, Invitrogen). Fluorescent
images were captured with a Nikon Eclipse TE300 Microscope. To correct
for background fluorescence, sections stained only with secondary
antibody were used, and corrected fluorescence intensity measurements
of MSC‐NICHE were normalized to control hydrogel devices at each
timepoint.
Islet Isolation
Allogeneic pancreatic islets were isolated from male Lewis rats
according to our previously published protocol.^[ [262]^16a,d ^]
Briefly, rats were euthanized with an isoflurane overdose immediately
prior to pancreas harvesting. The pancreatic duct was cannulated and
infused with 9 mL CIzyme RI collagenase (00 51030, Vitacyte) and
0.2 µg mL^−1 DNAse (dornase alfa; Genetech) dissolved in Hanks balanced
salt solution (HBSS; Gibco) supplemented with 10 mM HEPES (Gibco). The
pancreas was excised and enzymatically digested in a 37 °C water bath
for 19 min and 20 s, and the reaction was stopped with ice‐cold HBSS
containing 20% fetal bovine serum (FBS, Gibco), followed by mechanical
digestion. The digest was then washed three times with HBSS/HEPES,
filtered through a 500 µm mesh, and subjected to density gradient
separation using Optiprep (Sigma). Isolated islets were collected,
washed, and cultured in RPMI‐1640 media supplemented with 10% FBS,
20 mM HEPES, 5.5 mM glucose, 1 mM sodium pyruvate and 1%
penicillin/streptomycin (all from Gibco).
Islet Engraftment and Revascularization in NICHE Devices
MSC‐loaded NICHE devices were implanted in male Fisher rats. At 4 weeks
post‐implantation, diabetes was induced by a single i.p. injection of
STZ (50 mg kg^−1). After 5 weeks of vascularization, a subtherapeutic
dose (500 IEQ) of syngeneic pancreatic islets was transcutaneously
loaded into the NICHE cell reservoir of both Control‐No MSC rats (n =
6) and MSC co‐transplant rats (n = 6). At 1‐ and 4‐weeks
post‐transplant, tail vein injections of Lectin‐DyLight649 (1 mL at
1 mg mL^−1, Vector Laboratories) and Heparin (0.5 mL at 15 mg mL^−1, J.
T. Baker) were administered (n = 3 per group, per timepoint). Animals
were then euthanized by transcardiac perfusion with PBS followed by 4%
paraformaldehyde, and NICHE devices were explanted and fixed overnight.
Tissues were clarified and processed for insulin staining using the EZ
clear protocol.^[ [263]^21 ^] Briefly, tissues were delipidated, washed
and incubated with a 1:200 dilution of primary anti‐rat insulin
antibody (C27C9; Cell Signaling, 3014S) for 4 days. Next, the samples
were washed for 3 consecutive incubations of 2 h in PBS. The tissues
were then incubated with a 1:200 dilution of secondary antibody (goat
anti‐Rb AF555, Invitrogen, A‐21428) for 4 days. Finally, the tissues
were immersed in EZ view solution and incubated until equilibrated as
described in the referenced protocol.
Lightsheet Imaging and Analysis
Equilibrated samples were embedded in a 1% agarose hydrogel and mounted
on a custom sample holder.^[ [264]^21 ^] They were imaged in EZ view
solution using a Zeiss Lightsheet Z.1 microscope with a 5× lens at 0.5×
zoom. Tiled Z‐stacks with 20% overlap were acquired at a resolution of
1.829 µm × 1.829 µm × 7.03 µm (X:Y:Z). Lectin‐DyLight649 was excited
with a 638 nm laser at 10% power (300 ms exposure) and Insulin‐AF555
with a 561 nm laser at 5% power (30 ms exposure). The dataset was
stitched with Stitchy (Translucence Biosystems) and the generated 3D
images were analyzed using Imaris (Oxford Instrument) by a blinded
scientist. A threshold was applied to lectin and insulin positive
signal to obtain the volume of blood vessels and islets, respectively.
The portion of engrafted islets was calculated as the ratio of the
islet volume obtained via lightsheet analysis to the estimated volume
of transplanted islets, where 1 IEQ is equivalent to a sphere having a
diameter of 150 µm. Total vessel volume was determined by dividing the
blood vessel volume measured across nine ROIs per sample by the volume
of each ROI. Finally, the intra‐islet vessel volume was obtained by
dividing the blood vessel volume measured within the islets by the
total islet volume.
Allogeneic Immune Response Study in Diabetic Male and Female Rats
8‐week‐old Fisher male and female rats were rendered diabetic and
implanted with MSC‐loaded NICHE devices and insulin pellets as
described in subsection Implantation of NICHE in diabetic rats. After 5
weeks of vascularization, at day 0, rats were randomly assigned to one
of three groups: Islets + MSC, Islets‐only or vehicle. Rats receiving
allogeneic islets were transplanted according to their weight (15 000
IEQ kg^−1) and group co‐transplanted with MSC received syngeneic MSCs
at a 2:1 (islet: MSC ratio). Islets + MSC (n = 14 per sex) male rats
were co‐transplanted with 3600 IEQ and 2.7 ( 10^6 MSCs, and female rats
received 2200 IEQ and 1.65 ( 10^6 syngeneic MSCs embedded in a
thermosensitive collagen hydrogel (Advanced Biomatrix) and loaded in
two NICHE devices. Islets only (n = 14 per sex) rats received the same
amount of IEQ without MSCs loaded in two devices. Vehicle (n = 12 per
sex) served as control and were transcutaneously injected with collagen
hydrogel in the NICHE cell reservoir. All injections were performed
transcutaneously through the NICHE central silicon port with content
loaded in a 1 mL syringe equipped with a 22G x 1 needle as previously
described.^[ [265]^16a ^] BG was measured daily post‐transplant and
weight was monitored every other day. On day 3, control (n = 4 per
sex), Islets (n = 4 per sex), and Islets + MSC (n = 4 per sex) were
euthanized and both NICHE devices with surrounding tissue, and
peripheral tissues (spleen and draining lymph node) were collected for
analysis. At time points of days 7 and 14, control (n = 4 per sex),
Islets (n = 5 per sex), and Islets + MSC (n = 5 per sex) were
euthanized collecting same tissues for following analysis. Upon
excision, one NICHE device was processed immediately for CyTOF and the
other device was processed for histology and cytokine quantification.
Spleen and draining lymph node were processed immediately for flow
cytometry.
Imaging Mass Cytometry Analysis
IMC analysis was performed at the ImmunoMonitoring Core (Houston
Methodist Research Institute) using metal‐conjugated antibodies
prepared according to the Fluidigm protocol as previously described.^[
[266]^16 , [267]^51 ^] After epitope retrieval and blocking with 3% BSA
in TBS, slides were stained overnight at 4 °C with the antibody panel
in Supplementary table [268]1 and counterstained with Cell‐ID
Intercalator (Standard BioTools) before air‐drying and ablation with
the Hyperion system (Standard BioTools) for data acquisition. The IMC
data were preprocessed and checked for tissue integrity, staining
quality, and signal range prior to analysis. For every ROI, the single
cells are segmented using ilastik ^[ [269]^52 ^] and CellProfiler,^[
[270]^53 ^] based on DNA staining (Ir191) and other cell surface
markers. Following cell segmentation, mean intensities of each marker
for all single cells were extracted using the Histology topography
cytometry analysis toolbox (HistoCAT) ^[ [271]^54 ^] and data was
consolidated in R scripts for downstream analysis. The intensity values
for each marker were clipped at the 99.5 percentile to remove outliers
and normalized to a 0 to 1 scale to ensure equal weight across markers.
The normalized intensities were used for unsupervised clustering in
Seurat^[ [272]^55 ^] using Louvain algorithm.^[ [273]^56 ^] Cell
clusters were annotated based on the average expression of markers and
consolidated into 15 cell types. Cell densities for each type were
calculated by normalizing cell counts to the corresponding ROI areas.
Finally, Pearson correlation analysis was performed to compare the cell
density proportions between islet cells and the other defined cell
clusters. Analysis was performed on 1–2 ROI per sample with n = 3 for
each group in both sexes.
CyTOF Analysis
Single cell suspensions from NICHE cell reservoir tissue were obtained
via digestion with RPMI medium containing collagenase/hyaluronidase
(NC2031808, StemCell Technologies) with DNAse I (Roche, 100 µg mL^−1),
as previously described.^[ [274]^16a ^] Following digestion and red
blood cell lysis, the cell suspension was incubated with a metal‐tag
viability dye for 5 min, washed with cell staining buffer (Standard
BioTools), and subsequently stained for surface and intracellular
markers detailed in Table [275]S2, Supporting Information. Next, cells
were incubated with Cell ID Intercalator Ir (Standard BioTools) at 4 °C
overnight. The following day, cells were washed, and data was acquired
on the Helios instrument (Standard BioTools). Data were collected in
FCS files and analyzed with Cytobank, where normalization, filtering of
abnormal events,^[ [276]^57 ^] removal of beads and dead cells, and
gating on singlets and CD45^+ cells were performed. Finally, tSNE
analysis was conducted on the live CD45^+ singlets following the gating
strategy in Table [277]S3, Supporting Information, and the cell
population ratios were quantified accordingly. Finally, the heatmaps
displaying the fold change in cell abundance for the islets‐only and
islets + MSC groups relative to control were calculated and plotted
using R.
Cytokine Quantification
Concentrations of IL‐2, IL‐4, IL‐10, IL‐6, IL‐12p70, IFN‐γ, MCP‐1,
Fractalkine, and VEGF were simultaneously quantified in NICHE cell
reservoir tissue samples using a MILLIPLEX Rat Cytokine/Chemokine
Magnetic bead panel (Millipore, RECYTMAG‐65K). Tissues were extracted,
weighed and stored at −80 °C until homogenization with T‐PER buffer
(Thermo Scientific, 78 510) supplemented with Protease Inhibitor
Tablets (Thermo Scientific, A32955) (10 mL/gr of tissue), and protein
concentration was measured with the Pierce BCA Protein assay (Thermo
Scientific, 23 227). Prior to cytokine quantification, tissue
homogenate samples were centrifuged 10 000 × g for 10 min at 4 °C and
diluted 1:2 in the provided assay buffer. The plates were then setup
following the manufacturer's protocol, and sample fluorescence was
measured using a Luminex 200 reader (Luminex Corp). Analyte
concentrations were calculated by analysis of median intensity
fluorescence (MFI) data using a 5‐parameter logistic curve. The
calculated concentration values were normalized and expressed as
concentration relative to control.
TGF‐β quantification from NICHE tissue homogenates required incubation
with HCl followed by neutralization with NaOH prior to assay using the
Invitrogen Rat TGF beta 1 ELISA kit (Invitrogen, BMS623‐3), according
to the manufacturer's instructions. After absorbance measurements,
concentrations were determined using a 5‐parameter logistic curve and
normalized to the total protein content of each sample.
Flow Cytometry Analysis
Draining lymph node and spleen were collected for flow cytometry at
each timepoint. Lymph nodes were digested with
collagenase/hyaluronidase (StemCell Technologies) diluted 1:10 in
RPMI‐1640 for 1 min, then quenched with 2% FBS in PBS and filtered
through a 40 µm strainer. Spleens were dissociated by mechanical
filtration and subjected to ACK lysis (Quality Biological) to remove
red blood cells. Cells were washed, resuspended in 2% FBS in PBS, and
plated in 96‐well V‐bottom plates for staining. After blocking with an
FC blocker (ɑCD32, BD Biosciences) at 4 °C for 30 min, 1 × 10^6 cells
were stained with either a lymphoid panel (CD45, CD3, CD4, CD8a, CD25,
and viability dye) or a myeloid panel (CD45, CD11b/c, CD11b, CD80,
CD163, and viability dye) for 30 min at 4 °C. Following washes, cells
were fixed with eBioscience fixation/permeabilization buffer
(Invitrogen, 501 129 060) at room temperature for 20 min, permeabilized
with 1× permeabilization buffer (Invitrogen, 008 33356) and stained
intracellularly with Foxp3 (lymphoid panel) or CD68 (myeloid panel) at
4 °C for 30 min. Unstained and fluorescence minus one (FMO) samples
were processed in parallel. Cells were washed twice and re‐suspended in
PBS with 2% FBS. Data was collected on an A5SE instrument equipped with
FACSDiva v9 software (BD Biosciences) and analyzed with FlowJo v10
software (FlowJo, LCC) after gating out debris, doublets, and dead
cells. Treg population was defined as CD45^+CD3^+CD4^+CD25^+Foxp3^+.
Dendritic cells were defined as CD45^+CD11b^−CD11c^+ and macrophages
were defined as CD45^+CD11b^+CD68^+. Antibodies used for lymphoid and
myeloid panel are detailed in Table [278]S4, Supporting Information.
Gating strategy for lymphoid panel exemplified in Figure [279]S13,
Supporting Information for lymph node, and Figure [280]S14, Supporting
Information for spleen tissues. Gating strategy for myeloid panel
exemplified in Figure [281]S15, Supporting Information for lymph node,
and Figure [282]S16, Supporting Information for spleen tissues.
Visium HD Spatial Sequencing Library Preparation and Downstream Analysis
RNA quality assessment: RNA extraction was performed on 3 sections (5
µm each) from FFPE blocks of explanted NICHE devices from
allotransplant study to assess RNA integrity. Extraction was carried
out using the RNeasy FFPE kit for RNA extraction (Qiagen, 73 504) as
per manufacturer's instructions. RNA integrity was then measured by
(Agilent Technologies) using the High Sensitivity RNA ScreenTape
(Agilent Technologies, 5067–5579). Samples selected for sequencing had
a DV200 >30% along with positive DAPI stain in archival slides.
Visium HD spatial transcriptomics: 10 µm sections from FFPE blocks of
NICHE devices co‐transplanted with islets + MSC, explanted at days 3 (n
= 1 per gender) and 7 (n = 1 per gender) post‐transplant, were obtained
for Visium HD profiling according to 10× Genomics demonstrated protocol
([283]CG000408). Sequencing libraries were then prepared using the
Visium HD Reagent kits and Mouse Transcriptome v2 probes (10× Genomics,
1 000 674) following the 10× Genomics protocol ([284]CG000685). High
resolution H&E images were captured using a ZEISS upright microscope in
accordance with manufacturer's recommendations. The pooled libraries
were sequenced at 10× Genomics recommended depth using an Illumina
NovaSeq X Plus 25B PE150 sequencer at Novogene.
Downstream analysis: Primary data analysis was performed using
SpaceRanger version 3.1.2 (10× Genomics), including demultiplexing,
alignment, mapping and UMI counting. Specifically, for alignment and
mapping, the GRCm39‐2024‐A mouse reference genome and Visium Mouse
Transcriptome Probe Set v2.0 mm10‐2020‐A were used. The Space
Ranger‐generated clustering and projection cloupe files were imported
into Loupe Browser v8.0 (10× Genomics) for data visualization and
exploratory analysis. To better explore the NICHE local
microenvironment, pancreatic islet cell aggregates and their
surrounding tissue were manually selected based on H&E morphology.
Aggregated expression of pancreatic cell markers (Gcg, Ins1, Ins2,
Iapp, Ptprn, Mafa, NeuroD1, Pdx1, Pax4, Nkx2‐2) further validated the
accurate identification of islet cells. The selected ROI for all
samples (n = 4) were re‐clustered in Loupe Browser, obtaining 8–12 high
resolution clusters in each sample. The aggregated gene expression
matrixes for all clusters were extracted to Seurat R package^[ [285]^58
^] (v5.2.1) for downstream differential gene expression (DGE) and
pathway enrichment analysis.
To harmonize all identified clusters from all samples and identify key
genomic programs, tSNE analysis, followed by principal component
analysis (PCA) was performed using the gene expression matrixes. All 41
individual clusters were grouped into 3 main meta‐clusters based on
their similarity in gene expression. DESeq2^[ [286]^59 ^] was used to
perform differential gene expression (DGE) analysis by comparing the
gene expression between meta‐clusters. Upregulated genes associated to
each meta‐cluster were identified as signature genes and filtered by
‐log[10](p‐value) ≥ 1.5 and log[2]FC > 0.05 for upregulation.
Furthermore, gene set enrichment analysis (GSEA) was performed using
ranked gene lists from DGE analysis and computationally annotated their
functional pathways using clusterProfiler 4.0^[ [287]^60 ^] against the
Gene Ontology (GO) database. The most significantly regulated pathways
were identified using criteria of FDR < 0.1 and normalized enrichment
score (NES) > 1.
Quantitative Real‐Time Polymerase Chain Reaction
NICHE tissues were collected at day 7 post‐transplant from rats
receiving allogeneic islets only (n = 3 male and n = 5 female), and
rats co‐transplanted with MSCs (n = 4 male and n = 6 female). Collected
tissue was preserved in RNAprotect tissue reagent (Qiagen, 76 104) at
4 °C until homogenization using a Bead Mill 24 homogenizer
(Fisherbrand) and subsequent RNA isolation was performed following
manufacturer's instructions for RNA purification from animal tissues
using RNeasy Protect Mini Kit (Qiagen, 74 124). Isolated RNA
concentrations were quantified using NanoDrop One (Thermo Scientific)
and cDNA was generated with 2 µg of total RNA using the High‐Capacity
cDNA Reverse Transcription Kit (Applied Biosystems, 4 368 814). After
reverse transcription, cDNA was diluted 1:4 with Nuclease‐Free water
(Invitrogen, AM9937) and real‐time (RT) PCR was performed using TaqMan
Fast Advanced Master Mix (Applied Biosystems, 4 444 557) and TaqMan
gene expression assays (Table [288]S6, Supporting Information). All
RT‐PCR assays were performed with technical triplicates for all samples
using QuantStudio 6 Pro (Applied Biosystems). Gene expression levels
were normalized by the ΔΔCT method to housekeeping gene Gapdh and
relative expression was calculated as fold change (2^−ΔΔCT) in relation
to group receiving islets only.
Statistical Analysis
Results are expressed as mean ± standard deviation (SD) or standard
error mean (SEM) when deemed appropriate. Basic statistical analyses
were performed using Prism v.10 software (GraphPad Software Inc.). Mass
cytometry related statistical analysis were performed as mentioned in
Imaging mass cytometry (IMC) analysis and CyTOF analysis sections.
Spatial transcriptomics statistical analysis was performed as mentioned
in Visium HD spatial sequencing library preparation and downstream
analysis.
To compare means between two groups, we used two‐tailed Student's
t‐tests. For comparisons involving multiple groups, one‐ or two‐way
analysis of variance (ANOVA) was performed followed by post‐hoc
analyses. Specific analysis method, number of replicates, and p values
are specified in each figure legend. A statistically significant
difference was defined as p value < 0.05.
Conflict of Interest
S.C., M.F., C.Y.X.C., and A.G. are inventors of intellectual property
licensed by Continuity Biosciences. AG is a co‐founder and scientific
advisor of Continuity Biosciences. The other authors declare no
conflict of interest.
Author Contributions
J.N.C.C.: Conceptualization, Methodology, Validation, Formal analysis,
Investigation, Data curation, Visualization, Writing – Original Draft,
Writing – Review & Editing; S.C.: Methodology, Formal analysis,
Investigation; Data curation, Visualization, Writing – Review &
Editing; A.L.J.: Methodology, Investigation; N.H.: Methodology,
Investigation; T.B.: Investigation, Visualization; O.S.V.:
Investigation; Data curation; M.C.: Methodology, Investigation; L.F.:
Investigation, Visualization; M.F.: Data curation; G.E.R.:
Investigation; Y.X.: Methodology, Investigation, Validation; J.Z.: Data
curation, Formal analysis, Software, Methodology, Visualization;
L.B.A.: Data curation, Visualization; J.A.N.: Data curation, Formal
analysis, Visualization; F.N.: Investigation, Data curation; C.Y.X.C.:
Validation, Writing – Review & Editing; S.H.C.: Validation,
Supervision; J.E.N.: Validation, Supervision, Writing – Review &
Editing, Funding acquisition; N.S.K.: Validation, Supervision; A.G.:
Conceptualization, Validation, Writing – Review & Editing, Supervision,
Project administration, Funding acquisition.
Supporting information
Supporting Information
[289]ADVS-12-2411574-s001.docx^ (16MB, docx)
Acknowledgements