Correlative Light and Electron Microscopy (CLEM) Workflows

Correlative light and electron microscopy (CLEM) is a family of imaging workflows that combine the molecular specificity of fluorescence microscopy with the nanometer-scale ultrastructural resolution of electron microscopy (EM) on the same biological sample. The core problem CLEM solves is one that neither technique can solve alone: fluorescence microscopy can identify which structure or event you are looking at, using a labeled protein, organelle marker, or reporter, but it cannot resolve the surrounding ultrastructure at the nanometer scale. Electron microscopy can resolve that ultrastructure in exquisite detail, but on its own gives no molecular identity to any particular feature in the image. CLEM bridges the two by first locating a region of interest (ROI) by fluorescence, then relocating that exact same field in an EM dataset of the same sample.

The Core Principle: Locate by Fluorescence, Resolve by EM

The logic behind every CLEM workflow is the same regardless of the specific instruments or sample type involved: a fluorescently labeled feature (a tagged protein, an organelle marker, a reporter construct, or a rare transient event captured by live-cell imaging) is used to pinpoint a location or structure of interest. That location is then re-imaged, on the same physical sample, using EM to obtain ultrastructural context around it — membrane contacts, organelle morphology, macromolecular arrangement, or other detail well below the diffraction limit of light. The value of CLEM is specifically in that combination: it lets a researcher ask “what does the ultrastructure look like around this exact molecularly-defined event or structure,” rather than either guessing at identity from EM contrast alone, or accepting the resolution limits of fluorescence alone.

The General CLEM Workflow

Most CLEM protocols follow the same broad sequence, though specific steps vary by sample type (cultured cells, tissue sections, whole-mount specimens) and by whether room-temperature or cryogenic preservation is used:

1. Fluorescence imaging first. The sample is imaged by light/fluorescence microscopy — often live-cell imaging when the target is a dynamic or transient event — to identify and record the precise location of the region of interest. This step is what makes CLEM correlative rather than just “EM with extra steps”: it establishes molecular identity and, in live-cell applications, timing, before any EM-specific processing begins.
2. Sample processing and fixation compatible with both modalities. The sample must then be fixed and processed in a way that preserves both the fluorescence signal (or a route back to the same location) and the ultrastructure needed for EM. This is one of the central practical tensions in CLEM — see the challenges section below.
3. EM imaging of the same location. The processed sample is imaged by EM — conventional TEM, SEM, or, for cryo-preserved samples, cryo-electron tomography (cryo-ET) — targeting the same physical location identified in the fluorescence step.
4. Image registration and alignment. The fluorescence and EM datasets are registered against each other, so that the fluorescence signal (and whatever molecular identity or timing information it carries) can be overlaid onto the EM ultrastructure. Because the two imaging modalities produce very different contrast mechanisms, this registration is rarely a simple pixel-for-pixel overlay; it typically relies on shared reference points visible in both datasets.

Key Practical Challenges

Relocating the Exact Same Field of View

The hardest practical problem in CLEM is not imaging quality in either modality individually — it is finding the exact same location twice, on two different instruments, after the sample has been through processing steps (fixation, dehydration, resin embedding, sectioning, or vitrification) that change its appearance and sometimes its physical form entirely. Two general strategies are used to solve this:

• Fiducial markers — features visible in both fluorescence and EM images that serve as shared landmarks for registration. These can be intrinsic (a nuclear stain such as DAPI, which is visible by fluorescence and, with the right processing, produces contrast under EM) or extrinsic (fluorescent microspheres or nanoparticles added to the sample specifically to serve as correlation points).
• Grid-pattern references — for cultured cells or thin sections, gridded coverslips or finder grids etched with a coordinate pattern let a researcher record a ROI’s grid coordinate under the light microscope and relocate the identical coordinate under the electron microscope, independent of any biological fiducial.

Registration itself is typically done manually, by aligning shared landmarks by eye, or semi-automatically using dedicated correlation software; because EM and fluorescence contrast mechanisms are fundamentally different, fully automated pixel-based registration between the two modalities directly is generally not possible without an intermediate step such as a fiducial-based or machine-learning-assisted alignment.

Balancing Fixation and Labeling Protocols for Both Modalities

Standard EM sample preparation — strong aldehyde or glutaraldehyde fixation, heavy-metal staining, dehydration, and resin embedding — is optimized for ultrastructural preservation and contrast, but several of those steps quench fluorescent signal or degrade the epitopes needed for fluorescence detection. Conversely, fixation protocols optimized to preserve fluorescence (e.g., milder aldehyde fixation, avoiding certain dehydration or staining steps) can compromise the membrane and organelle ultrastructure that EM is meant to resolve. Most CLEM protocols represent a deliberate compromise between these two sets of requirements, and the exact balance point is sample- and question-dependent rather than fixed. Cryo-preservation workflows (rapid freezing rather than chemical fixation) sidestep some of this tension by avoiding chemical fixation and dehydration artifacts altogether, which is part of why correlative cryo-light microscopy paired with cryo-electron tomography has become an increasingly used variant for structural and cell-biology questions that require both molecular targeting and near-native ultrastructure.

Typical Use Cases

• Rare or transient cellular events. CLEM is particularly valuable when the structure or event of interest is too rare, too transient, or too heterogeneous to find by EM screening alone — a specific stage of cell division, a single instance of vesicle fusion, a viral entry event, or an autophagic structure at a defined stage. Fluorescence labeling identifies the rare event; EM then resolves its ultrastructure.
• Correlating dynamic live-cell fluorescence data with final ultrastructure. Live-cell imaging captures the dynamics of a labeled protein or organelle over time, but only fluorescence resolution. CLEM allows that same tracked event, once it reaches whatever biological end-point is of interest, to be fixed and imaged by EM — connecting a temporal, molecularly-defined observation to a static, high-resolution structural one.
• Target localization ahead of volume EM or cryo-ET acquisition. Because collecting a full electron tomogram or volume-EM dataset is time-intensive, CLEM is often used as a targeting step: fluorescence identifies where, within a large sample, the structure of interest actually sits, so that tomogram or serial-section acquisition can be focused there rather than searched for blind.

Related CASRAI Resources

For background on the EM side of a CLEM workflow, see Transmission Electron Microscopy: Sample Preparation Workflow and Resolution Limits, which covers the fixation, dehydration, and resin-embedding steps referenced above in more depth, and the Cryo-EM Microscope dictionary term for the cryo-preservation instrumentation used in cryo-CLEM variants. If a CLEM dataset includes cryo-EM or volume-EM data intended for public deposition, EMPIAR: The Electron Microscopy Public Image Archive for Raw Cryo-EM Data covers the relevant repository and submission requirements.

Frequently Asked Questions

What is the difference between CLEM and cryo-electron tomography (cryo-ET)?

They are complementary rather than competing techniques. Cryo-ET is an EM method for reconstructing 3D ultrastructure from a tilt series of a vitrified (rapidly frozen) sample. CLEM is the broader correlative strategy of using fluorescence to locate a region of interest before EM imaging. The two are frequently combined: fluorescence identifies where in a vitrified sample the structure of interest is located, and cryo-ET is then used to resolve that specific location in 3D — a workflow often referred to as cryo-CLEM.

Does CLEM require specialized fluorescent labels?

Not necessarily specialized labels beyond what a standard fluorescence experiment would already use (genetically encoded fluorescent proteins, antibody-based immunofluorescence, or small-molecule dyes), but the label and fixation protocol both need to survive, or be compatible with, the subsequent EM processing steps — which is the central practical constraint discussed above.

Can CLEM be done on tissue sections as well as cultured cells?

Yes. The core workflow is the same, though tissue applications generally add the complexity of first locating the ROI within a much larger, less uniform sample, and often rely more heavily on fiducial markers or serial-sectioning strategies to maintain registration between the fluorescence and EM datasets.