Protocol for visualization and quantitative analysis of murine lung immunological architecture using iBALT-UNMAP following bacterial challenge

Summary

The spatial distribution of adaptive immune cells is a critical determinant of the quality and magnitude of mucosal immunity. Here, we present a protocol for tyramide signal amplification (TSA) immunofluorescence that preserves the lung architecture and quantifies inducible bronchus-associated lymphoid tissue (iBALT) following murine Bordetella bronchiseptica infection. Starting from paraffin-embedded lungs, we describe steps for staining, imaging, and morphometric analysis. This protocol can be optimized for other pathogens and experimental settings to allow mapping of spatial immune responses within the respiratory mucosa.

For complete details on the use and execution of this protocol, please refer to First et al.¹

Graphical abstract

Highlights

• •Understanding lung immunological architecture is critical for interpreting immune responses
• •We established a tool to unify the nomenclature of lung B/T cell aggregates
• •Unbiased bioinformatic approaches to compare iBALT and aggregates across challenges
• •Defining lung immune architecture reveals unique host responses to specific pathogens

Publisher’s note: Undertaking any experimental protocol requires adherence to local institutional guidelines for laboratory safety and ethics.

The spatial distribution of adaptive immune cells is a critical determinant of the quality and magnitude of mucosal immunity. Here, we present a protocol for tyramide signal amplification (TSA) immunofluorescence that preserves the lung architecture and quantifies inducible bronchus-associated lymphoid tissue (iBALT) following murine Bordetella bronchiseptica infection. Starting from paraffin-embedded lungs, we describe steps for staining, imaging, and morphometric analysis. This protocol can be optimized for other pathogens and experimental settings to allow mapping of spatial immune responses within the respiratory mucosa.

Before you begin

Prior to conducting the experiments, ensure that the reagents are prepared according to the instructions provided in the materials and equipment section.

Innovation

This protocol builds upon standard immunofluorescence staining¹ and it transforms it into a powerful platform for spatiotemporal analysis of the entire lung immune architecture during bacterial infection. Beyond optimizing the experimental steps to reliably visualize inducible associated lymphoid tissue (iBALT), we introduce a standardized computational program, a code, capable of defining, quantifying, and comparing iBALT structures across experiments and pathogens using the staining protocols herein provided. Unlike current approaches, which rely primarily on B and T cell markers, our pipeline incorporates other cells markers, such as major basic protein (MBP) for eosinophils, an underappreciated but critical early driver of iBALT formation,¹ or GL7, a marker for germinal center B cells which correlates with the maturity stage of the iBLAT, to generate a more accurate and biologically meaningful representation of lung tertiary lymphoid structures.

Importantly, this work does not simply present a protocol, but establishes a comprehensive, workflow for spatial immunology of the lung. This includes image acquisition, multi-channel cell identification, structural annotation, and quantitative morphometric analysis. By pairing wet lab optimization with openly shared computational tools, this protocol offers a blueprint for standardizing how iBALT is defined, detected, and measured across research groups.

Overall, this protocol provides a foundation for studies aimed at characterizing lung mucosal immune responses during respiratory infection or sterile inflammation. Any project investigating iBALT formation or broader lung mucosal immune architecture can directly benefit from this approach. If widely adopted, this protocol and its accompanying computational pipeline have the potential to unify the operational definition of iBALT and promote greater reproducibility and comparability across studies.

Institutional permissions

All mouse procedures described this protocol were performed in accordance with IACUC regulatory standards and institutional guidelines (AUP:21-031, AUP:22-031, APU:24-036). Mice were originally purchased from Jackson Laboratories (Bar Harbor, ME) and subsequently bred in our facility. All animals were housed and cared for by the staff and veterinarians of the Animal Care Facility at Louisiana State University Health Sciences Center–Shreveport, which is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). Our mice were bred in-house to prevent changes due to differences in the mice microbiota. To prevent genetic drift, each mouse breeding colony was refreshed after every five generations. All mice were always maintained under a 12-h light/dark (7am–7pm) cycle in pathogen-free conditions under the care and maintenance of the institutional Animal Care Facilities.

A genotyping and experimental tracking database is maintained in the laboratory in both electronic and paper formats. This database records each individual mouse, including ear tag number, genotype, and the unique experimental code associated with all procedures performed, this includes infection type and days post-infection as well as any other procedure performed to the animal during this time. In parallel, staining standard operating procedures (SOPs) are maintained and coded. The SOPs correspond with a simplified protocol records that are archived electronically and in hard copy. Each SOP contains complete information for every staining performed, including sample identifier, incubation times, reagents, and antibodies used. Importantly, unique reference codes link animal records to their corresponding staining SOPs, enabling full cross-referencing and traceability of each individual mouse across experiments, procedures, and datasets. We recommend implementing a structured database and logging system that records genotype, age, and sex for every animal, ensuring accurate correlation with experimental outcomes. Such documentation is essential for identifying potential genotype, age, or sex dependent differences in downstream assays. Likewise, log forms should include all experimental calculations, including inoculum preparation, incubation times, and protocol details, to facilitate reproducibility and allow investigators to account for any variation observed among replicates.

All experiments are performed using at least two biological replicates, conducted once with males and once with females. If variability is high, biological replicates are increased to provide sufficient statistical power. Statistical analyses are first performed separately by sex; if no sex-dependent differences are detected, datasets are combined. When differences appear to trend but remain variable across samples, the number of biological replicates is further expanded to ensure robust statistical evaluation. Using the experimental models described in this protocol, we did not observe sex-dependent differences in eosinophil-driven mucosal immune responses following bacterial challenge.

Prepare an overnight liquid bacterial culture

Timing: 8–12 h

• 1.Inoculate a 15mL conical polystyrene tube containing approximately 5 mL of Luria-Bertani (LB)² or Stainer-Scholte (SS)^3,4 broth using an inoculation loop.
• 2.Allow liquid culture to incubate overnight (8–12 h) at 37°C, shaking at 200 rpm.
• 3.Measure the absorbance of each overnight culture at OD₆₀₀.
• 4.Based on the absorbance, assuming that an OD₆₀₀ reading of 0.1 is equivalent to 10⁸ CFU/mL of Bordetella spp., dilute the liquid culture with Phosphate-Buffered Saline (PBS) to achieve the desired inoculum concentration.^1,5,6,7

CRITICAL: For studies with other bacterial pathogens, adjust dilution amounts and use the recommended conditions for those specific pathogens.

• 5.Perform a series of 100-fold dilutions from the concentrated, prepared inoculum and plate these to enumerate CFU counts contained in the inoculums used to infect mice.⁶

Mouse infection with prepared bacterial inoculum

Timing: Variable

• 6.Anesthetize mice with 5% volume-to-volume (v/v) isoflurane using a gas-diffusion anesthetization chamber.

Note: The mouse is ready for inoculation once its respiratory rate has slowed to approximately 55–65 breaths per minute.

• 7.
  Using either overnight liquid cultures (see materials and equipment section) or prepared and frozen bacterial stocks that have been diluted to the right inoculum, inoculate each mouse nostril with 15–30 μL of PBS solution that contains a bacterial concentration of 10⁶ CFU/mL, for a total volume of 30–50 μL for each mouse.¹

  Note: 25 ul per nostril can also be used to reach a total of 50 ul of inoculum in the lungs of the mouse.

  Optional: To take further precautions, ensure no corneal reflex (blinking when touched) or limb movement by pinching the paws and tail periodically before proceeding.¹

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Hoechst 33342 (1:1000)	ThermoFisher	Cat. #H21486
BD OptiBuild™ RB744 Mouse Anti-Rat CD45R (1:50)	BD Biosciences	Cat. #757057
Purified anti-mouse CD3 Antibody (1:50)	BioLegend	Cat. #100201
Anti-mouse PNAd-AF647 (1:50)	BioLegend	Cat. #120807
Purified anti-mouse/human GL7 Antigen Antibody (1:50)	BioLegend	Cat. #144601
Anti-MBP/PGR2 (1:50)	Abexxa	Cat. #abx101775
Avidin, Alexa Fluor™ 488 conjugate (1:1000)	ThermoFisher	Cat. #A21370
Rabbit anti-Rat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 594 (1:1000)	ThermoFisher	Cat. #A-21211
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488 (1:1000)	ThermoFisher	Cat. #A-11008
Chicken anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647 (1:1000)	ThermoFisher	Cat. #A-21443
AF488 Anti-mouse CD3 Antibody (1:1000)	BioLegend	Cat. #100212
Bacterial and virus strains
Bordetella bronchiseptica RB50 (wildtype)	Cotter and Miller	Cotter and Miller8
Chemicals, peptides, and recombinant proteins
Xylene (mixture of isomers), Lab-Grade, VWR®	VWR	Cat. #77507-038
Ethanol absolute, pure (200 Proof)	VWR	Cat. #71001-754
Phosphate Buffered Saline (PBS)	Gibco	Cat. #10010049
Tween® 20 (Polysorbate) for synthesis, Sigma-Aldrich®	VWR	Cat. #EM8.22184.0500
Hydrogen peroxide (30 - 40%) 30% stabilised ACS analytical reagent, VWR Chemicals BDH®	VWR	Cat. #BDH7690-1
Fisher BioReagents™ Bovine Serum Albumin (BSA) Protease-free Powder	Fisher Scientific	Cat. #BP9703100
Normal Goat Serum	ThermoFisher	Cat. #PCN5000
ProLong™ Glass Antifade Mountant	ThermoFisher	Cat. #P36984
Citric acid, anhydrous for synthesis, Sigma-Aldrich®	VWR	Cat. #EM8.18707.1000
Sodium hydroxide, pellets ACS	VWR	Cat. #97064-476
Paraformaldehyde (PFA), Granular	Electron Microscopy Sciences	Cat. #19210
Isoflurane Inhalant Anesthetic	Attane	Cat. #RXISO250
70% Ethanol	VWR	Cat. #97064-768
Critical commercial assays
TSA amplification kit (Opal ™ reagent pack)	Akoya Biosciences	Cat. # FP1500001KT
Experimental models: Organisms/strains
Mouse:BALB/cJ Strain #000651	Jackson Laboratories	RRID:IMSR_JAX:000651
Software and algorithms
Keyence Image Analysis software	Keyence	–
Imaris 3D Analysis software	Imaris	Abu Bakar et al.9
GraphPad Prism (v10.6.1)	GraphPad	graphpad.com
https://doi.org/10.6084/m9.figshare.31236763	–	–
Other
VWR® Premium VistaVision™ Plain and Frosted Microscope Slides	VWR	Cat. #16004-422
VWR® Micro Cover Glasses, Rectangular, No. 2	VWR	Cat. #48382-126
Instantpot	Instant Pot ®	–
EasyDip™ Slide Staining System, Electron Microscopy Sciences	VWR	Cat. #102097-484
VWR® Gravity Convection Incubator	VWR	Cat. #89511-418
Microscope Slide Box, 100-Place	VWR	Cat. #77395-348
55S 11 × 14 Single Platform Laboratory Shaker	Reliable Scientific	Cat. #55
VWR® Disposable Plastic Transfer Pipette, Graduated	VWR	Cat. #10754-276
PAP Pen, Genetex	VWR	Cat. #89315-682
Chemical fume hood (Yurochko Lab)	–	–
Eppendorf™ Centrifuge 5430 R - Refrigerated Microcentrifuge	FisherScientific	Cat. #13-690-008
Keyence	Keyence	–
BD Syringes only with Luer-Lok™, Slip or Eccentric tips 3mL	VWR	Cat. #BD309656
BD Syringes only with Luer-Lok™, Slip or Eccentric tips 30mL	VWR	Cat. #BD302832
VWR® E Balances	VWR	Cat. #10204-990
VWR® Spinbar® Magnetic Stir Bars, Polygon without Pivot Ring	VWR	Cat. #58948-981
VWR® Advanced Magnetic Stirrers	VWR	Cat. #76447-052
Eppendorf Centrifuge, 5430R	Eppendorf	Cat. #97027-862
Vortex	VWR	Cat. #76416-196
Sharp Tip Dissecting Scissors (41/2″)	VWR	Cat. #82027-578
Waugh Forceps	VWR	Cat. #82027-428
Round-Bottom Polystyrene Test Tubes, 5 mL	Falcon	Cat. #352054
BD Nonsterile Syringes (5 mL)	VWR	Cat. #BD301027
Disposable Serological Pipettes (10 mL)	VWR	Cat. #75816-100
Disposable Serological Pipettes (25 mL)	VWR	Cat. #76320-280
Falcon Centrifuge Tubes, Conical Bottom (15 mL)	VWR	Cat. #21008-918
Falcon Centrifuge Tubes, Conical Bottom (50 mL)	VWR	Cat. #21008-940
Fisherbrand Microcentrifuge Tubes	FisherScientific	Cat. #14-666-318
Disposable Petri Dishes (100 × 15 mm)	VWR	Cat. #470191-208
VWR® Disposable Plastic Transfer Pipette, Graduated	VWR	Cat. #10754-264
Gilson™ PIPETMANTM 4-Pipette Kit	FisherScientific	Cat. #F167360G
Fisherbrand Pipette-Specific Tips 200	FisherScientific	Cat. #21-375D
Fisherbrand Pipette-Specific Tips 1000	FisherScientific	Cat. #21-375E

Materials and equipment

Citrate buffer (pH = 6)

Reagent	Amount	Final concentration
Citrate	1.92 g	–
NaOH, 10 M	1 mL	–
Tween-20	500 μL	–
ddH2O	1000 mL	N/A
Total	1 L	N/A

Store at 4°C for up to 1 month.

CRITICAL: NaOH and Tween-20 are highly dangerous chemicals that can burn or can be abrasive to skin upon contact. Disposable gloves must be worn when handling these reagents when preparing citrate buffer.

Phosphate-Buffered Saline with Triton-X-100 (PBS-T)

Reagent	Amount	Final concentration
1× PBS	1000 mL	–
0.5% Triton-X-100	500 μL	0.5%
Total	1 L	N/A

Store at 4°C for up to 1 week. Prepare fresh buffer between longer storage periods.

Note: It is recommended to add 4% Normal Goat Serum to PBS-T + BSA solution to limit non- specific binding of antibodies added later.

PBS-T + Bovine Serum Albumin (BSA)

Reagent	Amount	Final concentration
1× PBS	1000 mL	–
0.5% Triton-X-100	500 μL	0.5%
1% BSA	10 g	1%
Total	1 L	N/A

Store at 4°C for up to 1 week. Prepare fresh buffer between longer storage periods.

Step-by-step method details

Murine lung perfusion and fixation

Timing: 1 h

Here, we describe steps that are required to obtain and fix the lung for further examination with microscopy.¹⁰ We start with the perfusion and collection of the lungs, and describe all the steps up to the fixation.

Note: All this procedure needs to be performed in the fume hood.

• 1.Euthanize the mice with high dose of isoflurane and maintain a cotton pad embedded in isoflurane by the nose of the animal. Evaluate lack of reflexes prior to continuation.
• 2.Place mouse in dorsal recumbency and spray with 70% ethanol to reduce contamination.
• 3.Open the thoracic cavity carefully, avoiding contact with the gastrointestinal tract to limit contaminants.
• 4.Insert a needle into the left ventricle and perfuse 1 L of sterile cold PBS using a peristaltic pump at a recommended flow rate of 5–7mL/minute. If you have access to observed lungs, you might notice that lungs become pale.

Note: To perfuse use bottles of 250 mL of PBS and maintain cold in a bucket with ice while the mouse is perfused.

Note: In the absence of a peristaltic pump use a 50 mL syringe and very gently perfuse at least 50 mL of PBS. It is important to be extremely careful and had low flow to do not disrupt lung architecture.

CRITICAL: It is essential to maintain a steady hand to avoid breaking any tissue around the insertion area to prevent multiple needles slip out of the insertion site and reinsertion into the left ventricle. Doing this too many times will limit the perfusion from the tissues, as liquid will leak from the previous needle punctures in the heart.

• 5.Switch to 4% paraformaldehyde (PFA) and perfuse approximately 200 mL of cold PFA. For the last 50mL use a 30–50 mL syringe and inflate the lungs with the 4% PFA. Inflation of the lungs allow for the maintenance of the lung structure.

Note: To perfuse use bottles of 250 mL of 4% PFA and maintain cold in a bucket with ice while the mouse is perfused.

Note: During the PFA perfusion, there will be a moment that the tail goes up and slowly goes back down again. This is indicative that the tissues are fixed and ready to be extracted. Look for visual drainage of the blood from tissues and organs by observing the loss of red color from these areas.

Note: In the absence of a peristaltic pump use a 50 mL syringe and gently perfuse at least 50 mL of 4% PFA.

• 6.Dissect lungs carefully and placed them in a 50 mL Falcon tube containing 15 mL of 4% PFA.
• 7.Store at 4°C for 16–24 h.

CRITICAL: The maximum incubation time in 4% PFA should be restricted to 48 h.

Lung tissue fixation and processing for immunostaining

Timing: 3 days

Here, we describe steps that have to be followed to obtain the slides. Here we include the fixation, embedding, cutting and storage of the slides until staining.

• 8.After fixation in 4% PFA, replace with chilled 70% ethanol (4°C).
• 9.Store at 4°C until lungs are processed for up to 24–48 h.

Note: We recommend that lungs are not in 70% ethanol for longer than 48 h, as longer incubation times can damage the tissue, resulting in staining artifacts.

• 10.Process and embed in paraffin blocks with consistent orientation of lobes according to standard histology procedures^11,12
• 11.Cut 5 μm sections with the microtome mounting the sections on slides.

Note: We have processed the tissues in our histology on the core, but in case you have to do it yourself please follow the guidelines previously published^11,12

• 12.Incubate slides horizontally in a wet box at 60°C for 1 hour before long-term storage or staining. This enhances adhesion of the tissue to the slide, improves the flattening of the tissue, and minimize antigen degradation.

Deparaffination and rehydration of murine FFPE lung sections

Timing: 4 h

Timing: 30 min (for step 16)

Here, we describe steps for preparing samples for staining which includes the deparaffination and rehydration of the slides. This is very important to remove the hydrophobic paraffin wax and replace it with water allowing antibodies to penetrate and bind during the staining process.

• 13.
  Incubate slides with paraffin-embedded lung sections in a slide box at 60°C for ≤2 h.

  Note: If there is still paraffin visible after 2 h, incubate slides for an additional 30 min.

• 14.
  Remove slides from incubation and cool for 30 min at 27°C.

  CRITICAL: Ensure the vertical staining rack and slide staining jar are made of a xylene- and alcohol-resistant material, such as acetal polymer, prior to this step.

• 15.Insert slides into a vertical slide staining rack and begin a series of xylene washes for deparaffinization (Figures 1 and 2).
• 16.
  Submerge slides in 100 mL of 100% xylene on the Coplin or Staining Jar. Incubate for 5 minutes at room temperature.

  Optional: You can repeat steps 16 and 17 twice.

• 17.Remove slides from 100% xylene.
• 18.Submerge slides in 100% ethanol on the Coplin or Staining Jarr and incubate for 3 min at room temperature.
• 19.Replace solution with fresh 100% ethanol and incubate for 3 min at 27°C or room temperature.
• 20.Submerge slides in 95% ethanol in a Coplin or Staining Jarr and incubate for 3 min at 27°C or room temperature.
• 21.Replace solution to fresh 95% ethanol on the Coplin or Staining Jarr and incubate for 3 min at 27°C or room temperature.
• 22.Submerge slides in 70% ethanol on the Coplin or Staining Jarr and incubate for 3 min at 27°C or room temperature.
• 23.Submerge slides in 50% ethanol on the Coplin or Staining Jarr and incubate for 3 min at 27°C or room temperature.
• 24.Submerge slides in ddH₂O on the Coplin or Staining Jarr and incubate for 5 min while rocking at 60 rpm at 27°C or room temperature.
• 25.Repeat previous ddH₂O wash step twice.

Figure 1.

Lung tissue staining workflow

This schematic represents a simplified protocol for the staining, acquisition of images, and analysis of the images.

(A) It shows an schematic of the multiplex TSA staining.

(B) It shows the imaging and acquisition schematic of this protocol.

(C) It represents the data acquistion from the microscope to the computer where the analysis will be done using the unbiassed classification imaging code herein described.

Figure 2.

Murine lung tissue processing workflow

This schematic shows the simplified protocol for sample collection and processing including organ collection, sectioning, incubation times, and washes

Antigen retrieval of murine FFPE lung specimens

Timing: 1.5 h

Here, we describe steps that are required for antigen retrieval. This is a very important step to reveal epitopes that are undetectable due to the formalin fixation step.

• 26.Fill a pressure cooker about 1/2 with ddH₂O.

Note: Ensure the pressure cooker is filled with enough ddH₂O to entirely cover the bottom of the metal cooking bowl. For a 2-quart (1.8 L) cooker, add about 2 cups (0.5 L).

• 27.Fill the vertical slide staining jar with 80 mL of citrate buffer (Figure 3).
• 28.Preheat the container of citrate buffer for 15 mins using the “steam” pressure cooker setting.
• 29.Place the slides using a vertical slide rack in preheated citrate buffer and steam for 15 mins.
• 30.Leave slides to cool at 27°C for 15 mins.
• 31.Wash slides with ddH₂O 3× for 5 mins while rocking at room temperature, replacing with fresh ddH₂O after each incubation.
• 32.Wash slides with PBS-T for 5 mins while rocking.
• 33.Repeat previous step (PBS-T wash) 2×.

Figure 3.

Antigen retrieval

Schematic workflow representing the steps required for antigen retrieval including the incubation times, incubation method and description, and washes.

Blocking of non-specific binding sites and endogenous peroxidases

Timing: 1.5 h

Here, we describe steps that are important for the blocking of the samples. Blocking is a critical step to prevent unspecific binding of antibodies which can result in noise.

• 34.Using a hydrophobic barrier or PAP pen, draw a hydrophobic barrier around each tissue section and add blocking buffer directly onto the tissue within the barrier.
• 35.Incubate slides in 3% H₂O₂ diluted in PBS for 10–15 min at 27°C to quench any endogenous peroxidases.
• 36.Wash the slides for 5 min in PBS-T while rocking at 60 rpm. Repeat 2×, replacing with fresh PBS-T between each wash.
• 37.To ensure blocking of non-specific binding on the samples, incubate slides in blocking buffer (1% BSA + 4% NGS in PBS-T) at 27°C for 30–60 min, rocking at 60 rpm.
• 38.After incubation, remove any excess blocking buffer from each slide by gentle tapping. DO NOT WASH.

Tyramide signal amplification-based multiplex staining strategy

Timing: 3 days

The TSA staining (Figure 4) is used to amplify signal of low-abundant target and as such it is performed in three sequential cycles: Cycle 1 to stain for GL7 (germinal center B and T follicular helper (Tfh) cells) using Opal 520; Cycle 2 to stain for CD3 (T cells) using Opal 650, and Cycle 3 to stain for B220 (B cells) using Opal 570. Each cycle includes primary antibody binding, conjugated horseradish peroxidase antibody (HRP) secondary incubation, TSA development, and (for the first two cycles) HRP stripping. HRP stripping inactivates HRP from the stained slide, so that the same tissue section can be re-stained with another antibody^13,14,15

• 39.
  TSA Cycle 1 – GL7 (Timing: 10–12 hrs).

  • a.
    Place slides horizontally in a humidified slide box.
  • b.
    Using a PAP Pen, draw a hydrophobic barrier around each section.
  • c.
    Dilute GL7 primary antibody in blocking buffer (typically 1:100; optimize for your lot).
  • d.
    Pipette ∼100 μL of GL7 antibody solution onto each tissue section to fully cover the area.
  • e.
    Incubate overnight at 4°C in the dark.
  • f.
    Wash slides in PBS-T for 5 min at room temperature.
  • g.
    Replace the PBS-T and incubate 5 min at room temperature.
  • h.
    Repeat step 36 twice (total 3 washes).
  • i.
    Dilute goat anti-rat HRP in PBS-T or blocking buffer (typically 1:500).
  • j.
    Apply ∼100 μL to each section and incubate 15 min at room temperature.
  • k.
    Wash slides in PBS-T for 5 min.
  • l.
    Replace the PBS-T and incubate 5 min at room temperature.
  • m.
    Repeat steps k and l twice (total 3 washes) at room temperature.
  • n.
    Prepare Opal 520 working solution (1:100 in Tyramide Amplification Diluent).
  • o.
    Apply ∼100 μL Opal 520 to each section and incubate 10 min at room temperature in the dark.
  • p.
    Wash slides in PBS-T for 5 min at room temperature.
  • q.
    Replace the PBS-T and incubate 5 min at room temperature.
  • r.
    Repeat previous step twice for a total of 3 washes.
• 40.
  HRP Stripping after Cycle 1 (Timing: 1 h).

  • a.
    Prepare the Instant Pot with citrate buffer as described in the antigen retrieval section (fill the staining jar with citrate buffer and preheat on the Steam setting).
  • b.
    Place the slide rack into the preheated citrate buffer and steam on the Steam setting for 15 min.
  • c.
    Allow slides to cool at room temperature in the citrate buffer for 30 min.
  • d.
    Wash slides in distilled water for 5 min at room temperature.
  • e.
    Replace with fresh distilled water and incubate 5 min at room temperature.
  • f.
    Repeat previous wash step twice.
  • g.
    Wash slides in PBS-T for 5 min (3×).
  • h.
    Repeat blocking cycle (See Step 5: blocking of non-specific binding sites and endogenous peroxidases).
• 41.
  TSA Cycle 2 – CD3 (Timing: 2–3 h).

  • a.
    Dilute CD3 primary antibody in blocking buffer (1:200).
  • b.
    Apply ∼100 μL to each section and incubate 1–2 h at room temperature or overnight at 4 °C.
  • c.
    Wash slides in PBS-T for 5 min at room temperature.
  • d.
    Replace the PBS-T and incubate 5 min at room temperature.
  • e.
    Repeat step 59 twice.
  • f.
    Dilute goat anti-rabbit HRP (1:500) and apply ∼100 μL to each section.
  • g.
    Incubate 15 min at room temperature.
  • h.
    Wash slides in PBS-T for 5 min at room temperature.
  • i.
    Replace the PBS-T and incubate 5 min at room temperature.
  • j.
    Prepare Opal 650 working solution (1:100 in Tyramide Amplification Diluent).
  • k.
    Apply ∼100 μL Opal 650 to each section and incubate 10 min in the dark.
  • l.
    Wash slides in PBS-T for 5 min at room temperature.
  • m.
    Replace the PBS-T and incubate 5 min at room temperature.
  • n.
    Repeat previous washing step twice.
• 42.
  HRP Stripping after Cycle 2 (Timing: 1 h).

  • a.
    Prepare the Instant Pot with citrate buffer (see “HRP Stripping Cycle 1”) and preheat using the Steam setting.
  • b.
    Place the slide rack into the preheated citrate buffer and steam on the Steam setting for 15 min.
  • c.
    Allow slides to cool at room temperature in the citrate buffer for 30 min.
  • d.
    Wash slides in distilled water for 5 min at room temperature.
  • e.
    After this time replace with fresh distilled water and incubate 5 min at room temperature.
  • f.
    Repeat previous washing step twice.
  • g.
    Wash slides in PBS-T for 5 min 3×.
  • h.
    Repeat blocking cycle (See Step 5: blocking of non-specific binding sites and endogenous peroxidases).
• 43.
  TSA Cycle 3 – B220 (Timing: 2–3 h).

  • a.
    Dilute B220 primary antibody in blocking buffer (typical 1:200).
  • b.
    Apply ∼100 μL to each section and incubate 1–2 h at room temperature or overnight at 4 °C.
  • c.
    Wash slides in PBS-T for 5 min at room temperature.
  • d.
    After the incubation, replace PBS-T and incubate 5 min.
  • e.
    Repeat steps c and d twice.
  • f.
    Dilute goat anti-rat HRP (1:500) and apply ∼100 μL to each section.
  • g.
    Incubate 15 min at room temperature.
  • h.
    Wash slides in PBS-T for 5 min.
  • i.
    Replace the PBS-T and incubate 5 min at room temperature.
  • j.
    Repeat steps h and i twice.
  • k.
    Prepare Opal 570 working solution (1:100 in Amplification Diluent).
  • l.
    Apply ∼100 μL Opal 570 to each section and incubate 10 min in the dark.
  • m.
    Wash slides in PBS-T for 5 min at room temperature.
  • n.
    Replace the PBS-T and incubate 5 min at room temperature.
  • o.
    Repeat steps m and n twice.

CRITICAL: Do not perform HRP stripping after the final TSA cycle as this can weaken the signal.

• 44.
  Nuclear counterstain and mounting (Timing: 30 min).

  • a.
    Incubate slide with Hoechst 33342 (or a similar nuclear stain) diluted 1:1000 or 1:2000 for 30 min in PBS or blocking buffer. The precise dilution will vary according to the manufacturer recommendations.
  • b.
    Briefly rinse slides in PBS or distilled water to remove residual salts.
  • c.
    Gently wipe the back and edges of each slide with a Kim Wipe.
  • d.
    Add a small drop of ProLong Glass Antifade Mountant over the tissue.
  • e.
    Carefully place a coverslip to avoid air bubbles.
  • f.
    Allow slides to cure flat in the dark at room temperature for at least 8 h (overnight).
  • g.
    Slides are now ready for imaging.^13,16

Figure 4.

TSA multiplex staining

Schematic workflow representing the steps required for TSA multiplex staining. In this example we use first a staining using GL7, follwed by CD3 for T cells, and finally B220 for B cells. We add the colors used for each secondary to guide each cycle. Incubation times, washing, and stripping steps are included.

Imaging and analysis guidelines for characterizing murine iBALT

Timing: Variable

This section describes how to image and analyze murine lung sections stained as in Part A to identify inducible bronchus-associated lymphoid tissue (iBALT) based on B cell and T cell aggregates in peribronchial regions. Slides are multiplex stained for nuclei, B220, CD3, and GL7 using the sequential TSA-based protocol, then imaged on a widefield epifluorescence system for screening and on a spinning disk confocal microscope for quantitative 3D analysis in Imaris^17,18,19 A consistent, blinded workflow for field selection, acquisition, and batch analysis is critical for reproducible iBALT quantification (Figure 5).

• 45.
  Confirm staining and slide preparation.

  • a.
    Inspect slides briefly at low magnification to confirm presence of signal in all channels and absence of major artifacts (tears, folds, large bubbles).
  • b.
    Randomize and code slides before imaging so the operator is blinded to infection group and genotype.
  • c.
    Record slide codes and corresponding experimental information in a tracking spreadsheet.
• 46.
  Widefield epifluorescence imaging (Keyence).

  • a.
    On a Keyence BZX-800 microscope, use a low-magnification objective (4× or 10×) to identify bronchi and bronchioles.
  • b.
    Select peribronchial regions that contain lung parenchyma adjacent to the airway wall. Avoid fields with section folds, tears, or severe autofluorescence.
  • c.
    For each mouse, acquire at least 10 non-overlapping fields per section across three independent sections. We recommend using a 20× objective for optimal iBALT resolution and characterization. Include ∼6 mice per group and at least two independent experimental cohorts.
  • d.
    For each channel (Hoechst, B220, CD3, GL7), adjust exposure to maximize dynamic range without saturating bright pixels.
  • e.
    Save raw image files in a lossless format (for example, 16-bit TIFF) and organize them by experiment, mouse ID, section, and field number.

CRITICAL: All tissue slides must be imaged and analyzed using a blinded and unbiased approach during specimen staining, image acquisition, and quantification analysis.

• 47.
  Quantification of marker-positive cells in Keyence image analysis software.

  • a.
    Open the Keyence Image Analysis software and use the double extraction module to define detection parameters.
  • b.
    First, set a fluorescence intensity threshold for the target antibody channel (for example, B220), then set a threshold for the nuclear stain (Hoechst).
  • c.
    Define objects by nuclear staining and classify them as positive or negative for each antibody based on channel intensity.

    CRITICAL: Use the same threshold settings for all images within a given experimental batch to maintain consistency.

  • d.
    Run batch analysis for each group of images.
  • e.
    Export per-field measurements, including total Hoechst-positive nuclei and the number of nuclei positive for each antibody channel, as CSV or Excel files for downstream analysis.
  • f.
    Review a subset of images visually to confirm that thresholds capture true positive cells and exclude background.

    Note: If obvious over- or under-calling is observed, adjust thresholds and rerun the batch for that experiment.

• 48.
  High-resolution 3D imaging of peribronchial regions (spinning disk confocal).

  • a.
    Image the stained lung sections on an Olympus CSU-W1 spinning disk confocal microscope with a 40× objective.
  • b.
    Use laser lines and emission filters appropriate for Hoechst, B220, CD3, and GL7 with minimal spectral bleed-through.
  • c.
    For each mouse, select 2–3 sections and acquire 3–6 z-stacks per section in peribronchial regions that include airway epithelium and adjacent parenchyma.

    CRITICAL: Keep laser power, exposure time, detector gain, and z-step size constant within a batch.

  • d.
    Set the z-range to capture the full thickness of the tissue section, including the airway wall and peribronchial parenchyma.

    Note: Avoid saturated voxels and minimize photobleaching by limiting repeated scans.

  • e.
    Save image stacks as 3D multi-channel files (for example, OME-TIFF or Imaris-compatible formats) with filenames indicating experiment, mouse, section, and field ID.
• 49.
  Segmentation and surface generation in Imaris.

  • a.
    Open 3D image stacks in Imaris (version 10.2 or later). For each channel (Hoechst, B220, CD3, GL7), apply background subtraction to reduce low-frequency background.

    CRITICAL: Use the same filter parameters across all images within an experiment.

  • b.
    Create 3D Surfaces for the nuclear channel (Hoechst) using fixed parameters defined on a representative dataset.
  • c.
    Optimize smoothing, intensity threshold, and minimum surface volume so that individual nuclei are separated and closely apposed nuclei are not over-merged.
  • d.
    Generate 3D Surfaces for B220, CD3, and GL7 channels using thresholds and minimum volume settings that capture true lymphocyte structures while excluding noise.
  • e.
    Store these parameters as presets and apply them in batch mode to all images in the experiment.
  • f.
    Perform visual quality control for each image, starting with the nuclear surfaces.
  • g.
    Confirm that nuclei are well separated, that immune surfaces align with the underlying fluorescence, and that gross over- or under-segmentation is not present.

    Optional: If needed, rerun segmentation using the same preset parameters or apply minor adjustments and document any deviations.

• 50.
  Defining iBALT aggregates and assigning cell classes.

  • a.
    Using the 3D view, annotate peribronchial regions with high local nuclear density adjacent to the bronchial epithelium as aggregated (iBALT) and all other regions as non-aggregated parenchyma.

    CRITICAL: Perform this annotation blinded to group identity.

  • b.
    Assign nuclei to immune cell classes based on volumetric overlap with immune surfaces:

    • i.
      Nuclei that overlap B220 Surfaces are labeled B cells.
    • ii.
      Nuclei that overlap CD3 Surfaces are labeled T cells.
    • iii.
      Nuclei that overlap both GL7 and B220 or GL7 and CD3 Surfaces are labeled GL7⁺ B cells and GL7⁺ T cells, respectively.
  • c.
    Carry forward the aggregate status assigned at the nuclear level to each derived class.
  • d.
    For each cell class, distinguish cells located within annotated aggregates from those located outside aggregates in surrounding parenchyma.
• 51.
  Export of quantitative features and downstream analysis.

  • a.
    For each image, export summary statistics including the total number of nuclei, B cells, T cells, GL7⁺ B cells, and GL7⁺ T cells, stratified by aggregate versus non-aggregate compartments.
  • b.
    Export object-level tables containing per-cell features such as 3D coordinates, mean intensities in B220, CD3, and GL7 channels, and shape metrics (e.g., sphericity, ellipticity, area, volume) for nuclei and immune surfaces.
  • c.
    Save all exports as CSV files organized by experiment, mouse, and image.
  • d.
    Use these tables to compute per-mouse averages, compare iBALT composition across conditions, and generate figures in statistical software such as GraphPad Prism or R.

CRITICAL: Document all analysis parameters, including Imaris presets, thresholds, and inclusion criteria for aggregates, so that the workflow can be reproduced across experiments and laboratories.

Figure 5.

Analysis pipeline

Schematic workflow representing the analysis pipeline and classifications. Here includes the data export from the microscope computer to the analysis and graphic representation.

Spatial data analysis: iBALT-UNMAP (unsupervised mapping of iBALT architecture using iMAP)

Timing: Variable

This section describes the iBALT-UNMAP Microscopy Analysis Pipeline (iMAP), a Python²⁰ based workflow that uses 3D lymphocyte centroids exported from Imaris to call iBALT fields in an unbiased way. iMAP pools B cell and T cell centroids, clusters them with DBSCAN,²¹ builds 3D convex hulls around dense aggregates,^22,23 and applies fixed size and density thresholds²⁴ to classify images as iBALT, intermediate, or dispersed. Each lymphocyte is then labeled as inside or outside iBALT, which complements the composition and segmentation workflow described in Part B.

• 52.
  Before you begin the analysis, Install Python 3.10 with the following packages:

  • a.
    numpy²⁵
  • b.
    pandas.
  • c.
    scipy.
  • d.
    scikit-learn.
  • e.
    matplotlib (for optional plots).

Note: iMAP can be run in Jupyter Notebook or as a standard.py script.

• 53.
  From each 3D stack processed in Part B, export the following CSV tables from Imaris:

  • a.
    A nuclear “Spots” or “Surfaces” table with a unique object ID and centroid coordinates X, Y, Z (in microns).
  • b.
    Immune “Surfaces” tables for B220, CD3, and GL7 with object IDs and centroid coordinates.
  • c.
    A lineage or class label per nucleus if you already assigned B cell, T cell, and GL7 positive populations in Imaris.
• 54.Organize exports by experiment, mouse, and field in a consistent folder structure.
• 55.
  Use a base analysis directory structure for iBALT-UNMAP that contains at least:

  • a.
    Raw_Imaris_Exports.
  • b.
    Cleaned_Spatial_Data.
  • c.
    Cluster_Summaries.
  • d.
    Field_Calls.
  • e.
    Mouse_Summaries.
• 56.Set this base path at the top of the notebook or script.
• 57.
  Step 1: Import libraries and set global parameters.

  • a.
    Open the iBALT-UNMAP notebook (iMAP.ipynb) or script.
  • b.
    Import all required libraries at the top:

    import numpy as np

    import pandas as pd

    from scipy.spatial import ConvexHull

    from sklearn.cluster import DBSCAN
  • c.
    Define fixed parameters for the entire study:

    EPS = 25.0 # neighborhood radius in microns (example)

    MIN_SAMPLES = 10 # minimum lymphocytes per cluster

    MIN_CELLS = 50 # minimum cells for an iBALT cluster

    MIN_VOL = 5000.0 # minimum hull volume in μm^∧3 (example)

    Note: These values are chosen from pilot data and remain constant across mice, fields, and experiments.

    CRITICAL: Do not change eps or min_samples on individual images. That would introduce bias into the iBALT call.

• 58.
  Step 2: Load and clean Imaris exports.

  • a.
    For each image, read nuclear and immune CSVs from Raw_Imaris_Exports with pandas.read_csv.
  • b.
    Harmonize coordinate column names to x, y, and z in microns. Convert units if needed so that all tables use microns.
  • c.
    Merge nuclear and immune tables to create a per cell table with at least:

    • i.
      experiment_id.
    • ii.
      mouse_id.
    • iii.
      field_id or image_id.
    • iv.
      x, y, z coordinates.
    • v.
      lineage (for example B cell, T cell, GL7 positive B cell, GL7 positive T cell).
  • d.
    Filter this table to retain only lymphocytes used for clustering.
  • e.
    Include at minimum B cells (B220 positive) and T cells (CD3 positive).

    Note: You may keep GL7 positive populations as a separate column, but they are not required for cluster detection.

  • f.
    Save the cleaned per cell table for each image in Cleaned_Spatial_Data for traceability.
• 59.
  Step 3: Run DBSCAN clustering on B and T cell centroids.

  coords = cell_df[['x', 'y', 'z']].to_numpy()

  • a.
    For each image, extract B cell and T cell centroids into an N × 3 NumPy array:
  • b.
    Run DBSCAN with Euclidean distance.

    db = DBSCAN(eps=EPS, min_samples=MIN_SAMPLES)

    labels = db.fit_predict(coords)

    cell_df['dbscan_cluster'] = labels
  • c.
    Treat dbscan_cluster == -1 as noise (non-clustered cells). Values 0, 1, 2… define candidate lymphocyte aggregates.
  • d.
    Save the updated per cell table with DBSCAN labels back into Cleaned_Spatial_Data.
• 60.
  Step 4: Build convex hulls and compute cluster features.

  • a.
    For each image, iterate over distinct DBSCAN cluster IDs with label >= 0.
  • b.
    Extract coordinates for all cells in that cluster and compute a 3D convex hull:

    cluster_coords = coords[labels == cid]

    hull = ConvexHull(cluster_coords)
  • c.
    Record for each cluster:

    • i.
      cluster_id.
    • ii.
      n_cells (number of lymphocytes).
    • iii.
      hull_volume (e.g., hull.volume).
    • iv.
      hull_area (optional).
    • v.
      mean_nn_distance or similar density metrics (optional).
  • d.
    Store these per cluster metrics in a table and save it for each image in Cluster_Summaries.

    Note: Convex hulls approximate the 3D boundaries of lymphocyte aggregates and provide volume estimates used for iBALT thresholding.

• 61.
  Step 5: Apply predefined thresholds and classify clusters.

  • a.
    for each cluster, test whether it meets the pre-specified size and density criteria:

    • i.
      n_cells >= MIN_CELLS.
    • ii.
      hull_volume >= MIN_VOL.

      Note: You can optionally add a density threshold such as cells per μm³.

  • b.
    Assign a cluster level category:

    • i.
      iBALT_cluster if all thresholds are met.
    • ii.
      intermediate_cluster if clusters are present but do not meet full criteria.
    • iii.
      dispersed_cluster if below the cell count or volume threshold.
  • c.
    Store the cluster category in the cluster summary table and propagate it back to the corresponding cells in the per cell table.

    CRITICAL: Thresholds must be chosen before analyzing experimental groups and then held constant across all analyses.

• 62.
  Step 6: Label lymphocytes as inside versus outside iBALT.

  • a.
    For each image, identify clusters labeled iBALT_cluster.
  • b.
    For each such cluster, use its convex hull to define an “inside” region. A simple approach is to test each cell’s coordinates against the hull in Python.
  • c.
    For each lymphocyte, assign:

    • i.
      inside_iBALT = True if its centroid falls inside at least one iBALT hull.
    • ii.
      inside_iBALT = False otherwise.
  • d.
    Add inside_iBALT as a Boolean column to the per cell table.
  • e.
    Save the updated per cell table with DBSCAN label, cluster category, and inside versus outside flag.
• 63.
  Step 7: Call each field as iBALT, intermediate, or dispersed.

  • a.
    For each image (field), examine the set of cluster categories:

    • i.
      If at least one iBALT_cluster is present, label the field as iBALT.
    • ii.
      If one or more intermediate_cluster are present but no iBALT_cluster, label the field as intermediate.
    • iii.
      If only dispersed_cluster or no clusters are present, label the field as dispersed.
  • b.
    Create a field level summary table that includes:

    • i.
      experiment_id.
    • ii.
      mouse_id.
    • iii.
      field_id.
    • iv.
      field_call (iBALT, intermediate, or dispersed).
    • v.
      Number of clusters in each category.
    • vi.
      Total B cells and T cells in the field.
  • c.
    Save this table in the Field_Calls directory.
• 64.
  Step 8: Summarize per mouse and prepare for statistics.

  • a.
    From the per cell tables, compute mouse level summaries:

    • i.
      Total B cells, T cells, GL7 positive B cells, and GL7 positive T cells.
    • ii.
      Counts and fractions of each lineage inside versus outside iBALT.
    • iii.
      Mean and variance across fields for these quantities.
  • b.
    From the field call table, compute for each mouse:

    • i.
      Number of fields analyzed.
    • ii.
      Fraction of fields classified as iBALT, intermediate, or dispersed.
  • c.
    Save per mouse summary tables in Mouse_Summaries as CSV files for import into GraphPad Prism, R, or Python based statistics scripts.
  • d.
    Archive all parameter values (EPS, MIN_SAMPLES, MIN_CELLS, MIN_VOL), software versions, and directory paths in a README file inside the project folder.

Expected outcomes

Imaging and analysis of iBALT by multiplex TSA IF staining

Successful completion of part B will yield 1) high quality, multi-channel micrographs of the from FFPE lung sections with preserved airway and peribronchial architecture, 2) per field of view and per specimen/mouse counts of B220+ B cells, CD3+ T cells, and GL7+ lymphocytes, and 3) classification of fields of view as iBALT positive or iBALT negative based on organized peribronchial aggregates (Figure 6). Below we detail what should be observed for the staining herein described, we discuss the method of visualization, the recommended statistical analysis, and finally, the quality controls. It is important to note that we recommend including a negative staining control in each staining experiment. This control will be stained only with secondary antibodies.

Figure 6.

Example of lung with B/T cells forming iBALT versus lung with B/T cells recruited but not forming the characteristic iBALT

Depicted in white is the Horscht staining showing nuclei, in green is the CD3+ cells staining corresponding with T cells. Finally in red is B220+ cells which correspond with B cells.

For enumeration of B220⁺ B cells and CD3⁺ T cells (Figure 7), B220⁺ and CD3⁺ signals appear as discrete, cell-associated fluorescence with minimal diffuse background in well-stained sections. In bacterial challenge groups (relative to mock), you typically see increased peribronchial lymphocyte density and higher B220⁺ and CD3⁺ counts per field, with the effect magnitude depending on pathogen, dose, and timepoint. However, these results will vary depending on experimental design. For visualization, data is best plotted as the mean or sum of the cells counted for each mouse/specimen, not individual fields of view, which should be treated as technical replicates. Box/violin plots (or dot plots) for B220⁺ cells per field and CD3⁺ cells per field, where each dot is one mouse (mouse value = mean of its fields). Another option is to include stacked composition plot showing B cells vs T cells per mouse.

Figure 7.

Example of B/T cell quantification. Each point represents the number of B/T cells per Region Of Interest (ROI)

This symbol represents the number of B cells or T cells per region of interest, which correspond with the areas imaged around the peribronchial area. The barrs represent the standard deviation. Two-way ANOVA was used to determine statistical significance between groups.ns= non-significant, ∗ p<0.05; ∗∗ p<0.01; ∗∗∗ p<0.001; and ∗∗∗∗ p<0.0001

For manual classification of iBALT by peribronchial B/T aggregates, iBALT-positive fields show organized, dense peribronchial clusters containing both B220⁺ and CD3⁺ cells (often with a B cell enriched center and T cell enriched adjacent regions). iBALT-negative fields show scattered lymphocytes without dense organization near airways. For visualization per mouse, compute the Fraction of fields that are iBALT-positive (iBALT fields/total fields). You may also calculate the number of iBALT fields per mouse, but this is optional. After this, graph as dot plot (each dot = mouse), with group mean ± Confident Interval. For statistical analysis, analyze per-mouse proportions with Two-way ANOVA for proportions across multiple groups. For simpler comparisons (two groups), using Mann–Whitney on per-mouse proportions is acceptable if distribution is non-normal. If you want an independent readout of HEVs and tissue organization, we recommend adding PNAd to label HEVs or include a stromal marker panel for a validation marker. You can also use PNA as a germinal center–associated marker (depending on timing). Treat these as optional confirmatory stains, not required for the core protocol.

For manually classifying GL7-associated maturation of iBALT, GL7 signal should be localized to subsets of lymphocytes within organized aggregates. The biology is timepoint-dependent. A common pattern in maturing structures is enrichment of GL7⁺ cells within or adjacent to dense B220⁺ regions, but GL7 can label activated lymphocytes, so interpretation should be anchored to co-localization with B220 and/or CD3. For visualization, we recommend doing this per mouse and per plot. For each mouse, GL7⁺ B cells are defined as GL7 overlap with B220 (and nucleus), and reported as counts per field and/or fraction of B cells. GL7⁺ T cells are defined as GL7 overlap with CD3 (and nucleus). For each plot, graph a dot plot of % GL7⁺ of B220⁺ (per mouse), and a dot plot of % GL7⁺ of CD3⁺ (per mouse). Use a ratio metric (GL7⁺ B/total B) across timepoints if your study includes a time course.

To evaluate iBALT immunological landscape (expanded cell types and colocalization), additional lineages (CD4, CD8, eosinophils, APC markers, stromal markers) should exhibit structured compartmentalization relative to B/T zones and airway landmarks. Colocalization metrics should be interpreted as “spatial overlap at the voxel level,” not proof of cell–cell contact. For visualizing composition, use Dot plots (per mouse) of counts or densities of each lineage inside iBALT vs outside iBALT, or stacked bar charts showing mean fractional composition inside iBALT. For spatial relationships, graph the nearest-neighbor distance distributions (for example eosinophil-to-aggregate distance) summarized per mouse. For colocalization, graph using dot plots of Manders (or % colocalized area/voxels) per mouse, stratified by group.

iMAP obtained with the iBALT-UNMAP code (iBALT microscopy analysis pipeline)

Successful completion of this pipeline yields a per-cell table with x,y,z coordinates, lineage labels, DBSCAN cluster assignment, and inside_iBALT classification. From this, cluster summary tables with convex hull metrics (volume, cell count) and categorical calls. Perform a per-field call (iBALT, intermediate, dispersed) and per-mouse summaries of iBALT frequency and compartmentalization. To access the code of the iBALT-UNMAP, please access Figshare via https://doi.org/10.6084/m9.figshare.31236763. Coordinates should be in the same units (microns) across all images, DBSCAN parameters should be fixed across the full study, and iMAP calls should align with obvious aggregates in representative example images without tuning per image.

For code-based evaluation of iBALT with iMAP (Figure 8), you should expect iMAP returns DBSCAN-defined clusters and a field-level call (iBALT, intermediate, dispersed). In datasets with clear aggregates, iMAP should identify at least one high-density cluster meeting your fixed thresholds (MIN_CELLS, MIN_VOL) in iBALT-positive fields. For visualization, we recommend using agreements with operator calls for a confusion matrix (operator vs iMAP) across fields. Per mouse, plot fraction of fields called iBALT by iMAP. For cluster feature plots, distribute cluster sizes (n_cells) and hull volumes by group. Another optional scatter is hull_volume vs n_cells with clusters colored by category (iBALT vs intermediate).

Figure 8.

Example of a density plot obtained with the iBALT-UNMAP, our unsupervised mapping of iBALT architecture using iMAP

Red circles are shown B cells and green triangles are representative of T cells. Delineated black shapes are shown the aggregates https://doi.org/10.6084/m9.figshare.31236763.

Quantification and statistical analysis

All experiments should be performed in three independent biological replicates. The exact number of mice and technical replicates and the test performed must be indicated in each figure legend. Statistical tests should be applied based on previous literature^6,11 For animal experiments, we generally performed Two-Way ANOVA analysis. Similarly, all the data that is normally distributed must be analyzed using One-Way and Two-way ANOVA test and this needs to be indicated in each figure legend. When ANOVA is not an option, Student’s T tests should be performed. All graphs and data herein presented were analyzed using the GraphPad (v10.0.2). A p-value ≤ 0.01 was considered statistically significant. In the figures, the asterisks correspond with ∗ = p ≤ 0.05, ∗∗ = p ≤ 0.01, ∗∗∗ = p ≤ 0.001, ∗∗∗∗ = p ≤ 0.0001.

For analysis with the Keyence Image Analysis Software (v1.1.2) of each immunofluorescence assay, a fluorescence intensity threshold was set and used to detect Hoechst-stained cells that were also positive for each target antibody. This was done under the double extraction section of the software. Where the first step was to set the threshold for the antibody target and the second step was to set the threshold for the nuclear stain. Each group of images analyzed in this manner was done in batch analysis using the same threshold settings. We performed three independent sections of 6 different BALB/c mice per group (infected at different times) for staining and acquired at least 10 different images per section of each animal.

For statistical analysis of TSA multiplex IF staining, we recommend using Two-way ANOVA (factors: genotype, infection) on per-mouse means, followed by Sidak’s or Tukey’s multiple comparisons if comparing infection and genotype simultaneously. If there are only two groups, use unpaired t test (or Mann–Whitney if data is not normally distributed). As a quality control, confirm exposures are not saturating (avoid clipped signal), and confirm the same thresholding approach was applied across the batch (Keyence) and validated visually on a representative subset of fields.

For statistical analysis of manually classified GL7⁺-iBALT, use a Two-way ANOVA (genotype × infection) on per-mouse GL7 metrics. If multiple timepoints are included, use two-way or three-way ANOVA (genotype × infection × time), or mixed effects if repeated measures apply (usually they do not in terminal mouse timepoints).

For statistics of iBALT colocalization, use a Two-way ANOVA for per-mouse inside/outside metrics (you can analyze “inside” and “outside” separately or analyze a derived ratio: inside/(inside+outside). It is important to note that colocalization values often violate normality. Consider Mann–Whitney or transform or use mixed models if you retain per-field values but include mouse as random effect.

For statistics of iBALT-UNMAPs (iMAPs), use Cohen’s kappa (operator vs iMAP) for field-level agreements, or Two-way ANOVA (or nonparametric if needed) for mouse-level comparisons of iBALT frequency. To compare distributions of cluster features, test per-mouse summarized cluster metrics (for example, maximum cluster size per mouse, or mean of “iBALT clusters” per mouse).

Limitations

It is important to note that TSA multiplex staining is highly sensitive to antibody compatibility and retrieval/stripping conditions. Prior to performing the staining, confirm that all antibodies are suitable to tolerate repeated heat stripping. In addition, widefield screening of stained slides is efficient, but can be biased by field selection. Sample blinding and a fixed airway selection database is highly recommended to reduce this risk. It is also worth noting that iMAP performance depends on fixed parameter choices (eps, min_samples, MIN_CELLS, MIN_VOL). These must be selected on pilot data and held constant to avoid bias.

Troubleshooting

Problem 1: Tissue sections detach during antigen retrieval or TSA cycles

This may be due to insufficient slide adhesion, aggressive retrieval, or over-handling during jar transfers.

Potential solution

• •Use charged slides and ensured baking at 60°C for at least 1 hour (longer if sections are older).
• •Reduce mechanical stress: minimize rapid rack transfers and avoid banging racks on jar edges.
• •If persistent: shorten retrieval time slightly or switch to a gentler retrieval buffer for that antibody while keeping the stripping strategy consistent.

Problem 2: Weak or absent signal in one channel (B220, CD3, or GL7)

It may be that the primary antibody dilution is too high, insufficient antigen retrieval, expired/over-diluted Opal reagent, or HRP inactivation.

Potential solution

• •Verify the primary antibody on a known positive tissue first (single stain controls).
• •Titrate primary antibody (two to three dilutions around your current condition).
• •Confirm HRP secondary species match and that HRP is fresh and not cross-contaminated.
• •Prepare Opal working solution fresh and protect from light.

Problem 3: High background or autofluorescence that obscures true positives

This may be due to lung autofluorescence, incomplete blocking, inadequate washing, or nonspecific secondary binding.

Potential solution

• •Increase wash stringency (longer PBS-T washes, more changes).
• •Increase blocking time and consider adding an autofluorescence reduction step compatible with TSA (choose a method that does not quench your fluorophores).
• •Reduce exposure time and confirm you are not amplifying background by over-developing TSA (shorten TSA incubation slightly).

Problem 4: Channel bleed-through or apparent co-positivity that is not real

This may be due to overexposure, spectral overlap, or bright structures causing cross-channel detection.

Potential solution

• •Acquire single-color controls for each Opal channel to set exposure and confirm bleed-through.
• •Reduce exposure/laser power on the brightest channel.
• •Use sequential acquisition and appropriate emission filters. Avoid saturating pixels.

Problem 5: Inconsistent Keyence thresholding leads to inflated or deflated cell counts

This may be because thresholds are set on one “nice” image and applied blindly to images with different background or exposure.

Potential solution

• •Lock exposure settings during acquisition for a batch, then set thresholds on a representative set spanning low/high signal.
• •Validate thresholds by manual spot-checking 5 to 10 images per batch.
• •Export the same set of metrics every run and keep a “threshold log” with the exact values used.

Problem 6: Operator iBALT scoring is variable between people or drifts over time

This may be because there is no written scoring rubric, inconsistent field selection around airways, or due to subconscious unblinding.

Potential solution

• •
  Add a short scoring rubric:

  • 1) airway present, 2) peribronchial region included, 3) dense B/T aggregate present, 4) organized structure vs scattered cells.
• •Keep the operator blinded with a fixed coding spreadsheet.
• •Include “calibration fields” (a small set of example images scored at the start of each session).

Problem 7: Imaris segmentation over-merges nuclei or fragments surfaces (affects all downstream quantification)

This may be due to over-smoothing, thresholds being too permissive, minimum object sizes being too small, or uneven illumination.

Potential solution

• •Optimize nuclear surfaces first, then lock those settings as a preset.
• •Increase minimum object volume to suppress speckle noise.
• •Use background subtraction consistently across all images in a batch.
• •Perform a quick visual QC for every stack before export.

Problem 8: iMAP obtained with the iBALT-UNMAP returns no clusters

This may be due to coordinate units’ mismatch (pixels vs microns), too few lymphocytes retained after filtering, or eps/min_samples incompatible with your acquisition scale.

Potential solution

• •Confirm x,y,z are in microns for every file.
• •Confirm you are passing the intended cells (B and T) into DBSCAN.
• •Validate on a known iBALT-positive field and confirm lymphocyte count is sufficient.
• •If eps/min_samples must be changed, change them once using pilot data, then freeze them for the entire study and state that in the protocol.

Problem 9: iMAP obtained with the iBALT-UNMAP calls too many small clusters (false positives)

This may be because eps is too small, min_samples too low, or noise is not adequately filtered.

Potential solution

• •Increase min_samples or tighten the minimum cluster size threshold (MIN_CELLS) and minimum hull volume (MIN_VOL).
• •Add a density constraint (cells per volume) as an additional fixed threshold if needed.

Problem 10: Colocalization metrics look “great,” but images look wrong

This may be due to incorrect saturation, background overlap, or inappropriate thresholds inflate Manders/Pearson.

Potential solution

• •Always show the thresholder colocalization mask for representative images.
• •Use Manders (and % overlap voxels/area) with carefully documented thresholds rather than relying on Pearson alone.
• •Avoid drawing mechanistic conclusions from voxel overlap alone. Frame as spatial co-distribution.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Dr. Monica C. Gestal (monica.cartellegestal@lsuhs.edu).

Technical contact

Technical questions on executing this protocol should be directed to and will be answered by the technical contact, Dr. Nicholas J. First (first@uthscsa.edu).

Materials availability

This study did not generate new reagents. The code for the iBALT-UNMAP is available athttps://doi.org/10.6084/m9.figshare.31236763https://doi.org/10.6084/m9.figshare.31236763

Data and code availability

The code for the iBALT-UNMAP is availiable at http;//doi.org/10.6084/m9.figshare.31236763.

Acknowledgments

We would like to acknowledge the support of the funding bodies, which had no role in the study design, data collection, data analysis, or data interpretation: NIH/NIGMS Centers of Biomedical Research Excellence (COBRE) grant (P20GM134974), the Center of Excellence for Arthritis and Rheumatology intramural award, and the Intramural Research Council Seed and Start-up Package from LSU Health Sciences Center at Shreveport. We would also like to acknowledge the Predoctoral Multidisciplinary Training in Cardiovascular Pathophysiology (MTCP) Fellowship (5T32HL155022-03) of K.M.P. provided by the Center for Cardiovascular Diseases and Sciences (CCDS) at LSU Health Sciences Center at Shreveport and the Ken Peterson Memorial Predoctoral Fellowship of K.M.P. provided by the Department of Microbiology and Immunology at LSU Health Sciences Center at Shreveport. We would also like to acknowledge the Ike Muslow Predoctoral Fellowship of N.J.F. provided by LSU Health Sciences Center at Shreveport.

We would also like to acknowledge LSU Health Shreveport Research Core Facilities (RRID:SCR_024775), including FACS analysis (Flow Cytometry Core), RT-qPCR (Genomics Core), as well as Imaris usage and analysis for microscopy images (Microscopy Core). We would also like to acknowledge our CAIPP core facilities, established as part of our NIGMS COBRE funding (P20GM134974), including the CAIPP Bioinformatics and Modeling Core RRID:SCR_024779 and the CAIPP Immunophenotyping Core RRID:SCR_024781. We would also like to acknowledge the contribution of the Animal Models and Histology Core, which we used for embedding and sectioning (RRID:SCR 024776).

Author contributions

N.J.F., setup and optimization of protocols for performing the experiments described herein, establishment of microscopy criteria and analysis, developing the iBALT-UNMAP unbiased bioinformatic tool, and writing and editing of the manuscript; K.M.P., optimization of protocols, writing, editing, and formatting; J.P.-L. and J.M., writing and editing; M.C.G., setup and optimization of microscopy staining, writing and editing, funding, project administration, and supervision.

Declaration of interests

The authors declare no competing interests.