The European phama giant has agreed to pay Japan's Toyama Chemical up to $370m (€275m) for T-5224.
It prevents numerous genes being activated in response to inflammation triggers, including several genes whose protein products are the targets of current arthritis drugs.
The increasing trend towards licensing deals earlier in a drug's development is not necessarily down to pharma companies wanting it that way.
Although earlier does generally mean cheaper, the trend is at least partly down to the 'low-hanging branches' phenomenon - all the good late stage candidates have already gone and pharma companies are having to search harder and harder for innovative experimental drugs.
This is good news for the smaller companies who develop the drugs initially, as they reap the rewards and save considerable aditional clinical costs - it is now estimated to cost over $1.2bn to develop a drug.
Having completed preclinical research, Toyama initiated Phase I trials of T-5224 around a year ago.
In the case of T-5224, it blocks a specific transcription factor called Activator Protein-1 (AP-1).
This protein normally promotes the conversion of a gene's DNA into messenger RNA (mRNA) - the first step in protein synthesis.
By preventing AP-1 from working, several gene targets are inactivated 'downstream'.
"This novel oral compound complements Roche's developing portfolio of drug candidates in inflammation and rheumatoid arthritis," said Jean-Jacques Garaud, Head of Roche Pharma Development.
The company consider autoimmune diseases to be "one of the most important drivers for growth over the next few years" , with rheumatoid arthritis as the first indication.
The company has already launched MabThera (rituximab), the first and only selective B-cell therapy for rheumatoid arthritis.
Actemra is another novel therapy in their pipeline; it is a humanised monoclonal antibody to block the interleukin-6 (IL-6) receptor.
This target plays a major role in inflammation.
Also, ocrelizumab, a fully humanised anti-CD20 antibody, is just entering Phase III development for rheumatoid arthritis.
"The agent T-5224 has the potential to inhibit a key trigger of rheumatoid arthritis and has already shown promising pharmacological efficacy and safety in early clinical studies," continued Garaud.
AP-1 also plays another detrimental role in arthritis: in joint cells called osteoclasts, the molecule stimulates the production of enzymes that are thought specifically to cause the destruction of bones and joint tissue.
So, T-5224 has the potential to tackle the inflammation that characterises rheumatoid arthritis in more ways than one.
"By entering into a research and development collaboration with Roche, one of the world's leading research and development companies, we are able to increase Toyama's potential for novel drug development in the anti-inflammatory field, which is a field of concentration for Toyama Chemical," said Masuji Sugata, President of Toyama Chemical.
Yet, it is not just the small pharma companies who are looking to share development costs; this year Bristol-Myers Squibb (BMS) has penned two deals with AstraZeneca and Pfizer.
The first concerns two diabetes drug candidates and was worth up to $1.35bn.
The second deal , this time with Pfizer, was valued at potentially over $1bn and was over Apixaban, a direct factor Xa inhibitor designed to regulate thrombin and so prevent potentially fatal blood clots.
At the same time, BMS also agreed to pay Pfizer $50m in a co-development deal for a preclinical programme to treat metabolic disorders, including diabetes and obesity.
If big pharma is also getting in on the act, these early-phase deals look set to become even more commonplace.